1. Editorial
DON'T FORGET FROM 15th OCTOBER YOU MUST
USE THE NEW NUMBERS - EDITOR This is the first edition to be published under a new arrangement. Up until now our sponsors Perkin Elmer have funded the publication of IJVS from my office in the University of Southampton. Next week I hang up my boots - I'm retiring so my professorial office will close and we have decided to move off-campus. There are several very minor consequences as far as you the readers and contributors are concerned, of which the main ones are changes in contact numbers and addresses. My own editorial contacts remain the same as they always have been. Above are the new numbers, although they are headlined elsewhere in this edition. What effect will this have on your Journal - NONE! The staff, Louise Martin, our Webmaster and myself will continue exactly as before except that I will have a little more time to dedicate to the Journal! (better had !? - Assistant Editor) 100th Subscriber We have been keeping quiet about it but we were certain that during the Autumn of 1999, we would pass 1000 registered readers. The 1000th person to register was Libuse Pleserova from the Czech Republic, on 14th September . Wouldn't it be nice if we could reach 2000 next year! By the way, very few scientific journals have a circulation as big as ours - interesting thought! Couldn't be anything to do with the fact that IJVS is FREE? This Edition We don't normally talk much about future editions because somehow our plans always seems to come unstuck but we can announce that we will shortly produce an edition concentrating on Bio-Applications. Kai Griebenow has taken on the thankless task of editing it. Christmas and the Millennium New Year. We plan something light-hearted and different to mark the start of the year 2000, but NO I'm not telling you anything about it in advance! During 2000 we will definitely include an edition dedicated to Surface Enhanced Raman Spectroscopy and it's applications, to be edited by ZQ Tian at the University of Xiamen, China. I am also planning a Special Edition to bring you all up-to-speed on the calculation of vibrational frequencies and we will definitely highlight NIR during the year. Turning now to this edition - we have an excellent article from Ray Frost, Theo Kloprogge and Joelene Schmidt from Brisbane in Australia in which they show us all how valuable Raman is in the analysis of minerals. The subject isn't really new and Raman has a long way to go before it will really augment the standard methods, but Ray and his colleagues are obviously convinced it will be a valuable method in a few years time. Read the paper and tell us all what you think. I have a "bee in my bonnet" that the method we all used 30 years ago to present infrared data was one hell of a lot better than it is now! Oh no, I hear you all groan. When will the silly old duffer shut up? Have a look at what Anne de Paepe and I have to say and then send the blasts by e-mail. I always thought infrared emission spectroscopy was an ultra specialised method really of value only in the most academic research - an opinion based solely and exclusively on ignorance. I have never run an i.r. emission spectrum! Well it seems things have moved along and Drs Friedrich and Zahn want to tell us exactly how. They characterise textile fibres this way You will note that our notice board is really slim - Dear Reader has very little in it because you, the Dear Readers, are not sending in the goodies. I know you've all been at Conferences (on the beach!) but now we are all refreshed and bushy tailed get your keyboards going and send us some comments, questions and answers. The submitted section - three very varied offerings - an applied paper on Raman applied to ceramics, another on the analysis of cerium oxide films followed by a completely different subject - normal coordinate analysis. Dr Dodoff's paper reminded me that we really need a special edition on the calculation of vibrational frequencies. Editor
Ray Frost, Theo Kloprogge, and Jolene Schmidt
Introduction The identification of minerals is, in many cases, based on techniques like X-ray diffraction, optical microscopy and electron microprobe analysis. A major disadvantage of these techniques is that the mineral crystals have to be destroyed either to a powder or a thin section. Raman spectroscopy has been used for the identification of minerals [1-5]. This paper describes the application of Raman microscopy [6] as a non-destructive technique for the identification of minerals, suitable also for single crystals as small as a few hundred micrometers [7]. The identification of minerals generally starts in the field by visual examination by a
geologist. Based on properties such as colour, crystal habitus, hardness, lustre,
cleavage, etc. often in combination with detailed knowledge of the regional geological
history a first identification of many rock-forming minerals is possible. It becomes more
difficult for the more rare minerals mostly present in only minor amounts and often as
small crystals. In many of these cases visual examination even by an experienced
mineralogist does not give a definite answer and other analytical techniques are needed in
order to come to an identification of a mineral. The classical petrologist or mineralogist
will start with the preparation of thin sections of approximately 30 microns thickness of
the rocks sampled in the field for examination under the optical microscope. Optical
properties such as colour, pleochroism, refractive index, birefringes, etc. allows
minerals to be identified. However, this method is rather timeconsuming and depends
strongly on the experience of the geologist. Alternatively, minerals can be identified by
their crystal structure by X-ray diffraction and chemical composition by methods like
electron microprobe analysis. In general, X-ray diffraction uses powdered samples, whereas
the electron microprobe uses thin sections similar to those for optical microscopy, but
without coverglass and coated with carbon. This can pose a problem for the identification
of very small or rare crystals. An interesting non-destructive alternative is a
spectroscopic technique generally known as Raman microscopy. In this paper we report a
number of spectra of various minerals, all with crystals not larger than a few
millimetres. This will show the strength of the Raman microscopy technique as it will show
that each spectrum is unique and can be used as a sort of fingerprint for the
identification of the mineral when a large enough database of mineral spectra is
available. The Raman effect is a light scattering effect (see for example The Spectroscopists Bookshelf: http://www.ijvs.com/bookshelf.html). In order to obtain the Raman spectra of minerals, minerals were placed on a polished stainless steel surface on the stage of an Olympus BHSM microscope, equipped with 5x, 20x and 50x objective lenses. No sample preparation was needed. The microscope is part of a Renishaw 1000 Raman microscope system, which also includes a monochromator, a filter system and a charge coupled device (CCD). Raman spectra were excited by a Spectra-Physics model 127 HeNe laser (633 nm), recorded at a resolution of 2 cm-1 in sections of 1000 cm-1 for 633 nm excitation. Repeated acquisitions using the highest magnification, were accumulated to improve the signal to noise ratio in the spectra. For the 298 K spectra, data were collected at 20-second intervals for 10 minutes at maximum magnification. Spectra were calibrated using the 520.5 cm-1 line of a silicon wafer. It should also be noted that the filters in the Renishaw spectrometer start to eliminate the Rayleigh line at about 150 cm-1. This makes the determination of bands below 200 cm-1 difficult and without reliability. Spectra obtained with the microprobe normally consist of Raman bands superimposed on a background, which is a combination of the fluorescence and the instrument function. In the case of the HeNe laser and the measurement of the hydroxyl stretching region, the background is predominantly due to fluorescence and consists of a sloping linear baseline that is easily corrected. For the HeNe laser and the determination of the low frequency region, the background is predominantly a combination of the instrument function that is dependent on the behaviour of the notch filters, the grating-detector responses, and the fluorescent background. This background is induced by elements placed in the optical path between the sample and the entrance slit of the spectrometer. The background, although a complex function, is easily measured using the incandescent white light from the microscope. The measured spectra are then ratioed to this instrumental background to give the spectra as illustrated in the figures in this paper. Such a method ensures the correct instrumental function is taken into account and that the true spectrum is measured. Different instrumental profiles are obtained for the operation of different lasers. Normally with conventional dispersive Raman spectroscopy and to lesser extent, Fourier transform Raman spectroscopy data collection times can be excessively long, particularly for minerals. Often for poorly scattering minerals such as clays, several thousand FT Raman spectra must be coadded to obtain a spectrum of sufficient quality. In the case of Raman microprobe spectroscopy with multiplex detection, data collection times are very short and spectra are easily obtained within 1 to 10 minutes. Typically a collection time of 60 seconds, in which 10 spectra are co-added, is used with the maximum magnification. Power at the sample is in the 0.1 to 1 mW range. This use of such low power is a major advantage particularly over that used for FT Raman spectroscopy where typically the power used for minerals is 100 mW for a spot size of 200 m m. For Raman microscopy, the laser spot size is 0.8 microns and the laser power is between 1 and 4mW. Such low power means that there will be no heating effects and consequential damage to the mineral. It was found that the best quality spectra were obtained by scraping fresh crystals from a lump of the raw material on the metal support. Crystals of the kaolinite are easily observed under the microscope. One of the disadvantages of Raman microprobe analysis is the fluorescence of the clay minerals which is particularly pronounced when using the HeNe laser at 633 nm and other lasers of similar wavelengths. Each of the laser lines used to excite the Raman spectra of the kaolinite polymorphs caused fluorescence that subsided with time. Even so difficulty of measuring the low frequency region with the 633 nm excitation was experienced and the use of the diode laser operating at 780 nm did not alleviate the problem. Obtaining spectra at 77K, meant that the instrumental background shifted with temperature. This shift may be related to the change in the fluorescence of the sample at liquid nitrogen temperatures. Unique spectra from kaolinite polymorph crystals were more difficult to obtain from ground or powdered samples. If the crystals are smaller than the resolution of the microscope then spectra are obtained from an area that covers several crystals and an averaged spectrum is obtained. An advantage of Raman spectroscopy is that bands in the 200 to 400 cm-1 region are easily measured. Dispersive Raman spectroscopy inherently has the clay spectrum superimposed on the intense Rayleigh line. The use of Raman microscopy employing CCD detectors is greatly advantageous because of signal to noise improvement. Detectors in both FT Raman spectrometers and conventional dispersive instruments are photon noise limited. With a CCD detector this is no longer true: typically the noise from a CCD detector is at least 100 times less than from a semiconductor detector. The disadvantage of using Raman spectroscopy operating at the diode laser frequency of 780 nm is the loss of scattering efficiency. Raman scattering decreases as the fourth power of the wavelength of the scattered radiation, so the intensity of the Raman scattering at 780 nm compared to say 532 nm is ~7 times less. For the 633 nm (HeNe) laser, this factor is 2. Nevertheless, the disadvantage of this loss of efficiency is greatly outweighed by the advantages of the CCD detector. Further the advantage of using lasers operating in the 633 or 780 nm range rests with the reduction of the laser induced fluorescence which is a problem when using green or blue excitation wavelengths. Classification of minerals based on crystal-chemical composition and their Raman characteristics Minerals can be described as naturally occurring, inorganic crystalline materials. They can be classified based on their chemical characteristics. In general the following eight classes can be identified [8]:
Table 1. Characteristic frequency regions for various anionic groups. An important part of this classification is based on the occurrence of various characteristic anionic groups. Each of these groups is reflected as specific bands in both the infrared [9] and Raman spectra [10] and function as fingerprints for specific minerals. The characteristic frequency regions of the various characteristic groups are listed in Table 1. It has to be kept in mind that the following description is only very general and that in specific cases the situation can be much more complex. For example distortion of a symmetric anionic group in the crystal structure can result in a splitting of the Raman bands.
The sulphides are characterised by the metal-sulphur bonds giving characteristic bands in the low frequency region below 500 cm-1. Figure 1 shows the low frequency region of antimonite (Sb2S3), cobaltite (CoAsS), chalcopyrite (CuFeS2), enargite (Cu3AsS4), molybdenite (MoS2), kermesite (Sb2S2O) and the well-known pyrite or fools gold (FeS2). Although all minerals show the expected bands in the region below 500 cm-1, all minerals still show their own characteristic pattern useful for identification. Hydroxides give rise to hydroxyl stretching bands in the region between 3000 and 4000 cm-1, accompanied by bending modes around 600-1200 cm-1. The presence of water in the crystal structure is reflected in bands around 3200-3500 cm-1 and around 1600-1650 cm-1. The metal-oxygen bonds give rise to bands in mainly the low frequency region below 1200 cm-1. The exact position of the various metal-oxygen vibrations in the spectrum depends on factors like which metal is involved (atomic weight) and the bondlength and bond strength.
Figure 2 shows a few examples of oxide minerals like tantalite ((Fe2+,Mn)(Ta,Nb)2O6), anatase and brookite (both TiO2), ilmenite (FeTiO3), jacobsite ((Mn2+,Fe2+,Mg)(Fe3+,Mn3+)2O4) and the hydroxide brucite (Mg(OH)2). The carbonates around characterised by three major regions between approximately 1300-1550, 800-890 and 670-760 cm-1.
In Figure 3 the Raman spectra of some well-known and less well-known carbonate minerals are shown. These minerals include calcite (CaCO3), azurite (Cu3(CO3)2(OH)2), malachite (Cu2(CO3)(OH)2), gaudefroyite (Ca4(Mn3+)3-x(BO3)3(CO3)(OH,O)3), smithsonite (ZnCO3) and strontianite (SrCO3). Borates can be recognised by bands in the regions 1250-1350 cm-1 and 600-900 cm-1. Where borate bands are observed depends on the mineral structure and on the type of borate (metaborate BO22- or orthoborate BO33-).
Figure 4 shows the low frequency spectra of ulexite (NaCaB5O9.8H2O),
kernite (Na2B4O7.4H2O) and inderite (MgB3O3(OH)5.nH2O).
Ulexite and kernite show a typical borate band around 1007 and 880 cm-1. In
addition both kernite and inderite show two bands around 1350 and 1390 cm-1.
Nitrates show strong stretching modes around 1050 cm-1 accompanied by bending
modes in the regions between 800-850 and 715-770 cm-1. The sulphates are
characterised by strong bands in the region 900-1250 cm-1. Frequently they are
accompanied by a doublet in the region
Tungstates and molybdates show bands in approximately the same regions
around 750-950 cm-1 and 250-450 cm-1, whereas chromates show up in
the spectra only at the high frequency sides of these regions (800-950 and 350-500 cm-1).
Figure 6 shows the spectra of the tungstates scheelite (CaWO4) and wolframite
([Fe2+,Mn]WO4) and the molybdate wulfenite (PbMoO4). The
bands in the regions listed above are easily recognised. Between phosphates and arsenates
there are considerable similarities between their physical configurations of the
vibrations although they occur in different frequency regions (phosphate 950-1150, 920-970
and
Figure 7 shows the spectra of some selected arsenates, olivenite (Cu2(AsO4)(OH)), bayldonite (PbCu3(AsO4)2(OH)2), erythrine (Co3(AsO4)2.8H2O), annabergite (Ni3(AsO4)2.8H2O), scorodite (Fe3+AsO4.2H2O)and roselite (Ca2(Co, Mg)(AsO4)2.2H2O), and one arsenite (cafarsite (CaO)8(Ti, Fe)6O10(As2O3)6.2H2O). Strong arsenate bands are easily recognised in the region between 750 and 900 cm-1. This figure also shows the sharp difference between arsenates and arsenites. Vanadates have characteristic bands in the regions 700-900, around 870 and 300-400 cm-1. Figure 6 shows the spectrum of vanadinite (Pb5(VO4)3Cl) with a strong complex band around 800 cm-1 and multiple bands in the region between 275 and 400 cm-1. The silicates form the largest group of minerals on earth. The silicates can be divided in four major groups based on how the SiO4 units are linked together in the crystalstructure: ortho and ring silicates (e.g. olivine group, garnet group, tourmaline, beryl, etc.), chain silicates (e.g. pyroxene and amphibole groups), sheet silicates (e.g. clays and micas) and framework silicates (e.g. quartz, feldspars, zeolites, etc.). All silicates show very strong bands in the region between approximately 800 and 1200 cm-1, although the environment of the SiO4 group affects the shape and position of the various bands. Conclusions Raman microscopy has proved an extremely useful technique for the analysis of minerals. The use of the 633nm laser and the X50 objective proved the most useful combination. Fluorescence is not a problem for many minerals, and may be overcome through background subtraction. Impurities of lanthanides and actinides may cause this fluorescence which may prevent the collection of data. One of the difficulties is the collection of low intensity spectra on a large intense background. The collection of data for extended periods of time up to 30 minutes can overcome this difficulty. Nevertheless the advantages of the technique greatly outweigh the disadvantages, and the technique of Raman microscopy to the study of minerals offers a somewhat new and novel technique for the study of mineral chemistry. It can be said that Raman microscopy is a very powerful tool for the identification of minerals. It offers the possibility to identify minerals based on their specific Raman spectrum, which is unique for every mineral. In addition useful information is gathered about the presence of specific anionic groups and bonds like metal-oxygen or metal-hydroxyl bonds. Raman microscopy is a useful technique for the analysis of even very small crystals still present in handspecimens collected in the field, single crystals or even on uncovered thin sections like the ones used for electron microprobe analysis. Acknowledgments The financial and infra-structural support of the Queensland University of Technology, Centre for Instrumental and Developmental Chemistry is gratefully acknowledged. References
REF: Frost R., Kloprogge T., & Schmidt
Those of us who are too old to admit their age were bought up using continuous charts in our spectroscopy. Almost all the early infrared (or Raman) instruments output the data to commercial recorders made by such manufacturers as Honeywell Brown. These enormous devices generated a continuous ruled chart running over a horizontal roller. An ink pen (which clogged at the interesting moment) ran on a horizontal rail and hence a line (usually red) appeared on a long piece of chart paper. Some manufacturers built their own recorders into the instrument e.g. Grubb Parsons, but in the early days, the Honeywell Brown was king. Depending on scan speed and paper feed rate the wavelength or wavenumber scale varied but if a lab adopted a "house" set of conditions, the spectra were all roughly the same length. Of course, inter instrument comparisons were almost impossible. Clearly, the charts contained no grid to indicate wavelength or wavenumber, so we relied on the manufacturer generating a pulse (say every 50cm-1), we interpolated from these 'blips' and scribbled values on the chart. By the 1960's some order was taking over - an increasing number of manufacturers produced printed charts. Beckman and Perkin Elmer were using this solution as early as the late 1950's. Now the spectra appeared against a pair of scales and were directly comparable. Very rapidly, all the manufacturers fell into line and the quality of data presentation rapidly improved so that the charts could be filed, retrieved and compared at will. Some manufacturers divided the spectrum to keep the chart sizes handy, but in general all was logical and reasonable†. † Some manufacturers rolled the chart round a drum e.g. Perkin Elmer in the 21, 25 and 37 series instruments. Gradually, flat bed recorders became popular e.g. Beckman DK2 Visible - near infrared instrument, Unicam and Hilger Infrascan mid infrared instruments. Don't run away with the idea that once the pre-printed chart became available the use of the chart recorder was doomed. As late as the late 60's, chart recorder instruments were being produced, but the manufacturers overwhelmingly adopted the neater more compact and convenient pre-printed charts. Raman producers hung on longer using strip chart recorders well into the 70's. To give you some examples, we dug out some ancient spectra. Figure 1 is the Raman spectrum of solid K2CrO4. As you will see the frequencies are indicated by blips.
In Figure 2, we show an example of a spectrum recorded on a pre-printed chart. The spectrum is of Polystyrene and was recorded on a model 237 Perkin Elmer.
These days, we always present spectra as absorbance or transmission vs wavenumber, but this was not always the case. Until the 60's, many infrared spectra were recorded as transmission vs wavelength. In the States, this presentation lasted longer than it did in Europe. Comparison between these infrared spectra and their Raman compliments was, of course, quite impossible and comparing a linear wavenumber spectrum with a linear cm-1 one was a real teaser. In Figure 3, we show an example of a wavelength spectrum. Note that the 3000cm-1region is horribly compressed whilst the low frequency range is spread out. Remember 3000cm-1 is 3.33µ but 750cm-1 approximates to 13µ.
Early FTIRs output their data to printers and the format was consistent e.g. the Nicolet MX-1 and the Perkin Elmer PP1 printer. Both systems produced long charts with sensible displays and the absisca was logical and predictable. As a result, if a 'house' presentation was adopted spectra were easily comparable. Then came the PC and logic went out of the window - spectra were presented on A4 (or US standard) sheets and both the x and y scales could be to any scale the user intended or just finished up with. The net result is that you cannot visually compare spectra. You all know to your cost what we mean, but in Figure 4 we illustrate the point; the spectra are presented in just sufficiently different ways to make visual comparison a nightmare.
Does it matter when the data is on the hard disk? Yes! When we talk to
our colleagues, weI need diagrams - not all the spectra we have ever recorded, but only
the really useful ones. We need with colleagues to compare spectra so the presentation
must be comparable. Dozens of spectra on slightly different axes are a menace. Some tell
us the answer is to use the computer, but this is no solution. It is worth setting up a
formal presentation on the computer, but the cost in time of collecting the potentially
valuable data on the computer is too time consuming for informal frequent meetings and
discussions. Further, presentation of data on the computer can be very hard to read unless
projection is used. No - paper is still invaluable so the presentation should be
comparable spectrum to spectrum and instrument to instrument including i.r. to Raman.
You will see that the spectrum is divided so that it fits easily on an A4 (or American equivalent) page. The use of a grid spaced at a logical pitch in mm's (10cm-1 per mm)means that if no frequencies are indicated you can use your eye to estimate the cm-1 value and/or confirm it with a ruler. My proposal is that we the users of spectrometers - no less than 1000 of us debate what format is ideal and then we try very hard to get it adopted Internationally as a standard display. Back when pre-printed charts were used, those who wanted to present high resolution data or only wanted to consider a small part of the spectrum broke with standard format. I see no reason why this could not be the case in my proposal. So please answer the following questions and e-mail your replies to the Editorial Office. 1. Do you agree that a "standard" presentation format would be valuable both in-house and in comparing printed data between labs? 2. Do you like the format proposed in Figure 5? If you have your doubts, please tell us more. If you don't like our suggestions please send an alternative. 3. Where infrared and Raman spectra are to be overprinted together on the same diagram, is it appropriate to use two differently coloured lines? Hence - 4. Do you use a colour printer? Are colour printer facilities widely available in your organisation? 5. Do you prefer having the band cm-1 values printed directly onto the chart or do you prefer a table (remembering how easy it is to estimate the wavenumber value within the format we suggest)?
4.
Infrared Emission Spectroscopy Marion Friedrich and Dietrich R.
T. Zahn Email: friedrich@physik.tu-chemnitz.de Introduction Infrared emission spectroscopy is a commonly known method for the detection of radiation from stars distant from our planet. It is, however, still rarely applied to characterise solid samples on earth. This may be due to the fact that the experiment involves some difficulties which are related to the choice of a suitable reference and the precision of temperature measurements. To date emission spectroscopy investigations have been carried out with samples which are also well suited for conventional reflection or transmission measurements. Such samples are transparent KBr disks [1,2] or thin coatings on metals [2,3,4-7]. However, it is much easier to record absorption spectra than emission ones leading to the rare use of emission spectroscopy. In this study we present an introduction to infrared emission spectroscopy, discuss possible sources of error which may influence the measurements and demonstrate the suitability of emission spectroscopy for the characterisation of textile fibres, i.e. samples without any well-defined sample geometry. Such fibres are thus not suitable for transmission spectroscopy. Of course, it would be possible to prepare KBr pellets, but pellet preparation requires a lot of time and may also be difficult for this kind of material. First of all, we briefly introduce the relevant formulae, terms and facts which are fundamental to the understanding of emission techniques. Basics of Emission Spectroscopy The most important laws describing the emission of radiation are Kirchhoff“s law [7] and Planck“s formula [8]. According to Kirchhoff“s law, the amount of absorbed and emitted
radiation is equal for any body in thermal equilibrium with the environment. If a body is
a strong absorber of radiation at a certain wavenumber
with Figure 1 shows the spectral energy density distribution for a black body at different temperatures. With increasing temperature the magnitude of the emitted radiation increases and the maximum shifts to higher wavenumbers. For emission spectroscopy of textile fibres the temperature range from room temperature up to approximately 450K is of interest since the sample temperature must be held below the melting temperature of fibres. From the radiation distribution curves it is obvious that the upper wavenumber limit for the detection of emission spectra increases with the sample temperature. The working temperature of the typical infrared source, i.e. the globar, is at or slightly below 1200K, hence the full range of the mid infrared can be accessed easily. Figure 1 emphasises that we have to measure a signal in emission spectroscopy which is much lower in value than the source signal in absorption spectroscopy and as pointed out above the full frequency range may not be available.
The simplest sample which is as well suited for both
emission as for absorption measurements is a flat disk or a substrate covered with a thin
film. If such a sample is irradiated by infrared light, the light beam is partly reflected
at the surfaces and interface, partly transmitted and partly absorbed by the sample.
Neglecting scattering effects the absorptance
with
The experimental confirmation that equation (3) is valid is presented in figure 2 for a boron nitride film on steel with an optical thickness of 8.6µm. For this film reflectance and emittance measurements were taken. The boron nitride sample is particularly well suited to explain the measurement results. First of all, it is clearly visible that the emittance and reflectance spectra are complementary. Without any measurement error the sum of both should be exactly equal 1 at the same temperature. In the case of our samples some error may be introduced through the fact that the steel substrate was not perfectly polished and that we may not have measured reflection and emission spectra exactly in the same manner. Furthermore, we observe in the emission spectra measured with a DTGS detector a significant increase in noise with increasing wavenumber. As mentioned above the reason is the drastic decrease of the magnitude of the radiation towards higher wavenumber (see Figure 1) characteristic for the low sample temperature used in this experiment [of about 360K]. From both the emittance and reflectance spectra we can conclude that the boron nitride film is hexagonal since very strong features at 800cm-1 and 1400cm-1 marked by in Figure 2 by horizontal lines occur in the spectra. Finally interferences which are useful for the determination of film thickness can be seen in the spectral range above 1700cm-1.
We will now take a closer look at the measurement procedures and some important factors affecting the final result. In the emission experiment we measure the intensity of the radiati on radiating from our samples. The resulting so called single beam or single channel spectra contain the information about the optical properties of the emitting sample and its temperature. However, single beam spectra are seriously influenced by the optical design of the spectrometer, the detector used and background radiation, for example from the beamsplitter in the spectrometer. These factors which affect the lineshape of emission spectra as measured by an interferometer are thoroughly described in references [11] and [12]. There an expression is also given for the detected intensity which is valid if the detector temperature is well below the sample temperature T
where The method mainly used requires measurements of sample spectra
This experimental method is then well suited if an exact temperature
measurement can be performed and a suitable black body reference is available. The
advantage is that we do not need to know any special characteristics of our equipment as
are described by the instrument response function Complementarily, an interesting method is presented in [14] together with some calculations for the investigation of thin liquid samples using a thick opaque sample of the same liquid as reference instead of a black body, but this is beyond the scope of this paper. An alternative method is to determine the instrument response function
for the experimental setup once and for all. Therefore we substitute in (5) the difference
of black body single channel spectra by the product of an instrument response function Applying equation (3) to substitute also the emittance we get
Equation (6) can now be used to determine the response function via
reflectance and emission measurement of a reference sample. To minimize errors the
reference sample should be a black body. Having the instrument response function
As the emittance depends on the temperature, the positions of all characteristic sample features usually shift towards lower wavenumbers with increasing temperature. Furthermore, characteristic features broaden with temperature. At extremely high temperatures the spectra become blurred and look black body-like. For this reason it is important that the temperature difference is not too high if difference methods are applied. The question arises whether the infrared source in the spectrometer may serve as a reference and consequently make such a complicated procedure obsolete. Indeed it is in principle an almost perfect reference. However, we need to know precisely its radiation characteristic and temperature. Figure 3 shows the response function of the spectrometer used with a DTGS detector and determined using equation 5. Together with this function the single channel spectrum of the empty spectrometer with globar divided by the Planck function for a source temperature of 1150 K is presented. If the globar would show a black body like behaviour the resulting line shape should be identical to that of the calculated response function. The emittance spectrum of the globar, however, shows two pronounced dips at 470cm-1 and 1100cm-1. These dips on both sides of the Reststrahlenbande of SiC (at 780cm-1) are caused by differences in the globar surface reflectivity. It would be a step forward for a widespread application of emission spectroscopy if the globar emittance curve would be available for taking qualitative measurements routinely with computer assisted black body correction. Due to ageing effects and source temperature variations other methods as the determination of the response function are clearly preferable for quantitative measurements.
Experimental Determination of instrument response function We have used a thick non-transmitting pellet of KBr and graphite as reference since its reflectance is much lower than the reflectance of pure graphite [16]. Best results were found for pellets with a medium reflectance between 5% and 6%. Attempts to further reduce the reflectance by decreasing the graphite content of pellets were not successful, because the pellets became more and more transparent, in particular in the lower wavenumber region. For such samples equation (2) becomes invalid. The resulting instrument response function for the DTGS detector is presented in Figure 3. With this response function the emittance of a flat sample with a large
enough area can be calculated using equation (7) where For quantitative evaluation all measurements have to be done with the same laser frequency or optical velocity of the interferometer mirror as used for the determination of the instrument response function. If this is impossible intensities must be corrected since the detector signal magnitude depends on the measurement conditions. The boron nitride spectrum in Figure 2 is measured applying the procedure described above. In comparison to the DTGS the MCT detector shows a signal if the sample is held at room temperature. This different behaviour of detectors is caused by the different working temperatures. While the DTGS is in thermal equilibrium with the environment, the cooled MCT detector measures sample and background radiation due to its lower temperature. Owing to the compensation of background radiation terms in the difference of single beam spectra a similar response function can also be derived for different graphite sample temperatures with equation (6) using the MCT detector. In this case we have to consider for the calculation of the Planck function that the detector counts photons. Characterisation of textile fibres Using the sample holder for fibre samples the calculated emittance spectrum deviates by a factor from the true emittance spectrum. This is caused by differences in the emitting sample areas and their optical projections on the detector area. However, the spectra still reveal only characteristic sample features. Furthermore, some error may occur as a result of filament emission. According to equation (3) the emission of the metallic tantalum filament alone is very low because metals have a very high reflectivity. Therefore the contribution of the wire can be neglected in most applications. But a higher emission can be observed after formation of an oxide layer on the wire. Another effect which may become important is well known from emission spectroscopy of gases. Self absorption can be observed if the sample temperature decreases locally in the direction towards the detector. The worst case is presented in Figure 4. Spectra 1 and 2 are taken from polyethylene strips with a white painting. The polyethylene strips were wound around the heating wire 2 times. In the first spectrum (black line) the strips are in contact with each other. The second spectrum (red line) shows the result in the case of getting loose. The upper part of the strip is cooler and absorbs the radiation emitted from the strip underneath. We therefore observe dips especially in the main polyethylene peaks at 730cm-1 and 1465cm-1. This effect becomes important whenever absorption is strong as indicated by the difference spectrum 3 in Figure 4. It should be mentioned that fibres wound around the filament only in a single layer yield the best results because commercial textile fibres are thick enough for emission spectroscopy. Moreover, in most cases thinner fibres should result in better spectra.
Now, after discussing sources of error we present some spectra in order to demonstrate that emission spectroscopy is a perfectly appropriate tool for fibre characterisation.
Figure 5 shows the emittance spectrum of a polyester fibre wound around the heating wire and the reverse transmittance spectrum of a 6µm thick polyethylene terephthalate foil. The polyester fibre emission spectrum shows all characteristic features of polyethylene terephthalate, as can be seen by comparison with the foil spectrum. Small deviations are observed in the intensity ratio for example of the features at 845cm-1 and 875cm-1 marked by arrows. In addition, those features are more dominant in the emission spectra, which are weak in transmission, as a result of a higher effective thickness of emitting material. The dips in the strongest features at 1100cm-1, 1260cm-1 and 1725cm-1 of the emission spectrum are more likely caused by reflection [14] at the fibre-air interface than by self absorption due to a temperature gradient in the sample.
Figure 6a shows single beam spectra (corrected by the instrument response function) of commercially available aramide fibres with trade names Kevlar and Twaron. The Kevlar spectrum corresponds well to the micro-emission spectrum of the same material measured by DeBlase and Compton utilising a microscope with MCT detector [8]. Both fibers are fabricated from p-aramide and show therefore the same features. The influence of black body correction with the difference of Planck functions on the Twaron spectrum is shown in Figure 6b. After correction the features at higher wavenumbers become clearer and more pronounced. Since it is difficult to determine the exact temperature of the fibres we have calculated the emittance spectrum for the most probable temperature and temperatures deviating by ± 10° . All three spectra are well suited for further evaluation. As demonstrated by Figures 7a and 7b we can clearly distinguish by means of emittance spectra between the p-aramide fibres and the m-aramide fibre with trade name Nomex. The spectra measured with the DTGS detector are represented by red lines. The blue curves are emittance spectra from the same samples but measured with the MCT detector. For the same material we observe the same features independent on the detector. The spectral range of the MCT spectrum is limited by the cut off frequency.
Summary We have shown that emission spectra are as characteristic for the sample material as transmission spectra. In the case of fibres, emission spectroscopy has the advantage, that it does not demand any special sample preparation as known from transmission spectroscopy. Good emission spectra of fibres can be observed using a heated wire as sample holder. It should be useful to collect emission spectra of fibres in a library. Even without such a library a simple comparison with the reverse transmission spectrum as shown in Figure 5 should also yield good results in fibre identification. References
REF: Dietrich M. & Zahn D.R.T. Internet J. Vib. Spec.[www.ijvs.com] 3, 4, 4(1999) |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(reciprocal wavelength) it is also a
strong emitter at that frequency. Its emittance 
and absorptance 
are equal at any temperature T. Emittance and absorptance
characteristically depend on the type of sample and its geometry. The emittance 
is defined as the ratio of
emitted energy of the sample to that of a black body. A black body is a sample which
absorbs all incident radiation. Hence from Kirchoff at any temperature, no surface can
emit more radiation than a black body i.e. the black body source has the maximum possible
emittance 
. As a
perfect emitter the black body serves as a reference standard for the radiative properties
of a sample. Its spectral energy density distribution 
is determined by Planck“s function [3,9,10] 
(1).jpg)
(2)
:
transmittance, 
:
reflectance. Now as we have seen under the same geometrical and thermal conditions the
absorptance and emittance of any sample are equal whence Equation (2) for a non-transparent
sample reduces to 
(3).jpg)
(4)
= instrument
response function; 
=radiation intensity of
black body; 
= background radiation; 
background radiation reflected off the sample
towards the detector; 
=reflectance of the
sample. In practice it is impossible to avoid the measurement of additional background
radiation
and black body reference
spectra 
at two
different temperatures [11,13]. From these four measurements the sample emittance can be
determined to be
(5)
in equation (4). 
and the difference of
Planck functions at equivalent temperatures. Since the Planck function is calculated
instead of measuring the radiation intensity of a black body the derived response function
deviates by a factor including the geometry of experiment from the response function used
in equation (4). 
(6)
. .jpg)
for the
Bruker IFS 66 spectrometer used equipped with an emission port. Corresponding to equation
(6) it is necessary to take single beam emission spectra at known temperatures for a
reference sample under the same optical conditions as used for the emittance determination
via reflectance measurement. The ideal reference sample should have both a high emittance
and low reflectance and should not show characteristic features. Since the absorptance and
also the emittance of any sample depend slightly on the temperature [15] the procedure is
only applicable in a limited narrow temperature range. 
may be room temperature, .jpg)
