# Far-Infrared Fourier Spectroscopy as a Method for Structure Determination in Chemistry.

код для вставкиСкачатьFar-Infrared Fourier Spectroscopy as a Method for Structure Determination in Chemistry New analytical methods (4) By Erich Knozinger[*I Couched in the often confusing term "Fourier Spectroscopy" is the combined exploitation of a fundamental physical concept, a simple mathematical principle, and use of modern data technology. Introduction of this relatively young method into instrument technology has given great impetus to nuclear magnetic resonance and infrared spectroscopy. The infrared spectroscopic investigation of low frequency molecular vibrations has especially profited : in this connection one speaks of far infrared (FIR) Fourier Spectroscopy, which has opened up numerous interesting applications in organic, inorganic, and physical chemistry. 1. Introduction Nowadays, the methods ofvibrational spectroscopy---IRand Raman spectroscopy-are used along with NMR and mass spectroscopy for the elucidation of chemical structures". 411. In principle they afford an almost complete overview of the vibrational behavior of molecules, from which structural parameters can be determined qualitatively and quantitatively. The two methods are distinguished by the complementary nature of the information which they provide. Vibrations accompanied by a strong change of dipole moment can be observed particularly well in the IR spectrum (intense IR absorption bands); vibrations accompanied by a strong change in polarizability are readily observed in the Raman spectrum (intense Raman lines). If a molecule has a center of symmetry the vibrations are observable either in the IR spectrum only (exclusively IR-active vibrations) or in the Raman spectrum only (exclusively Raman-active vibrations). However, the ideal case for vibration-spectroscopic analysis, in which the frequencies of all normal vibrations of a molecule can be determined, seldom occurs. The reasons for this are far less theoretical than practical ones-equipment parameters and methods of sample preparation effect reduction in information quality, particularly in the region of low frequency vibrations. In the first place more or less pronounced difficulties are met withdepending on the measuring principle in the IR spectroscopic investigation of low frequency vibrations-due to the smallness of the signals to be measured. In this connection the Fourier method or Fourier spectroscopy as a measuring principle has enabled considerable advances to be made over the conventional dispersive methods. One refers to the spectral region of electromagnetic radiation, which on the one hand embraces the low frequency vibrations and on the other is a region in which the advantages of the Fourier method become particularly apparent, as the far IR or FIR region. Its upper and lower limits lie at C N . 500 c m - ' (20pm; 1 . 5 ~ 10'" Hz) and IOcm-' (1000pm: 3.3 x 10" Hz). respectively. In the present review the.types of vibrations to be expected in the FIR region are first briefly characterized. The wealth of structural information which they disclose seems to justify the considerable outlay on apparatus necessary for making the FIR region accessible. Currently one already has the choice of half a dozen commercially available FIR apparatus. They all function, with one exception, on the same measuring prin~~ [*I ciple: the Fourier method. In order to provide a fully illustrative account of this measuring principle and its more important consequences the following review will deal with the most striking advantages of the Fourier method in the FIR region as well as the fundamental principles that it has in common with conventional dispersive methods. 2. Applications of FIR Spectroscopy in Chemistry What is the information that can be derived from the FIR spectrum of a specific substance? It comprises information about the rotational and vibrational behavior of molecules. In this field the main accent is laid on vibrations since only a very few small molecules such as H 2 0 , D 2 0 , H2S, CO. NH3, HCI, etc.". show pure rotational transitions in the FIR region. Apart from these exceptions the typical spectral region for pure rotational transitions is the microwave region. (For gaseous samples rotational transitions can of course be ~uperimposed[~] on vibrational transitions.) If we assume a harmonic oscillator the frequency 0 (measured in cm l ) of the vibration of a diatomic molecule can be calculated from the relationship: where m, is the reduced mass, k the force constant, and c the velocity of light. For polyatomic molecules the one vibration ofthediatomic molecule is replaced by a series of so-called normal vibrations. In general the more complicated is the structure of a molecule-and here we include giant molecules such as exist in the form of crystals-the more complicated becomes the calculation of the frequencies and types of its vibrations. Equation (1) can be generalized for polyatomic molecules as follows: The frequency or the, wave number of an individual normal vibration is the smaller. the greater the masses (for rotatory vibrations moments of inertia) concerned in the vibration and the smaller the forces (for rotatory vibrations torques) that oppose the displacement of individual molecular components from the equilibrium positions. On the basis of several decades of experience[*] with the assignment of MIR spectra (MIR region: 4000 to CU. 400cm-') of organic and inorganic substances one can extrapolate from this generalization of relation (1) to the FIR region. One can thus predict . . Dr E. Knoringer Lehrstuhl fiir Physikalische Chemie der Gesamthochschule 59 Siegen 21, Postfach 210209 (Gcrmany) [*] T h e first absorption spectra in the I R region were meastired by W I*. Cohlrnrz in 1905. 25 which molecules will undergo normal vibrations with frequencies in the FIR region. The relevant types of vibration will be characterized in the following sections and explained by means of examples. The numerical values quoted for the force constants should be regarded as only approximate. In the present progress report it will not be possible to deal with the numerous applications of FIR spectroscopy in the field of physics particularly solid-state physics. For this purpose the reader is referred to the book by M d k r and Rorli,schi/d'21which contains, inter alia, a review of all present-day physical applications of FIR spectroscopy. 2.1. Stretching Vibrations Normal vibrations are described as stretching vibrations if they lead predominantly to changes in the bond lengths. This type of vibration is observed in the FIR region['] if heavy constituents are present in the molecule and their movement actually contributes to the change in the relevant bond length. In this connection the heavy constituents are considered--naturally only as a rule-to be the elements of the fourth and higher periods of the Periodic Table. Typical examples of application are the FIR investigations of carbonyl-["] and halo-metal complexes[' '. "]. The probability of finding a definite stretching vibration in the FIR region is further increased by two effects: - not only the masses of the atoms but also their ionic radii increase from period to period. The larger the ionic radii of the bonded atoms, the longer is the bond length and, according to Badger's rule18],the smaller is the force constant. - it is often not only the masses of the atoms directly involved in a particular chemical bond but also larger masses that are effective, as is shown when under certain conditions the bonded atoms vibrate approximately in phase with the ligands attached to them. In such cases, the term skeletal vibrations is used (Section 2.3). 2.3. Skeletal Vibrations of Macromolecules The skeleton of macromolecules consists in general of relatively light atoms, e. g. carbon atoms, but it undergoes forms of vibration in which large groups of neighboring skeletal atoms vibrating in phase vibrate in opposition to other such groups. The force constant is then approximately of the same magnitudeas in the stretching or bending vibrations of skeletal atoms of a comparable smaller, e. g. triatomic, molecule; however, in general very much larger masses are concerned, as they are-to a first approximation-the sums of the masses of atoms vibrating in phase. Correspondingly, the wave numbers for such skeletal vibrations are low. A special case of these skeletal vibrations of macromolecules is provided by the so-called one-dimensional lattice vibrations of long-chain unbranched molecules. These include, for example, the "accordion vibration" of unbranched aliphatic hydrocarbon^['^^ "1, where the wave number is smaller than 220cm-' when ten or more carbon atoms are concerned; admittedly, this vibration is Raman-active only when the chain skeleton consists wholly of the same type of atoms (e.g. carbons) and when the substituents on the skeletal atoms are all of the same type. 2.4. Skeletal Vibrations of Ringahaped Molecules The skeletal vibrations of ring-shaped molecules generally occur in the FIR region if there are only negligible changes in the bond lengths and bond angles. Such vibrations are observed in the case of planar or almost planar rings, if the ring atoms swing out of the "ring plane". In the case of four- and five-membered rings this form of vibration is known as ring-puckering vibration (cf. Fig. 1). /& '* 26 / I I/ I i 2.2. Bending Vibrations Bending vibrations is the term applied to normal vibrations in which mainly the bond angles change. They occur at lower wave numbers than the comparable stretching vibrations, since the forces that oppose displacement of molecular components from their equilibrium positions are always stronger along the direction of the bond than perpendicular to this direction. Thus, bending vibrations are found in the FIR region for all compounds whose stretching vibrations are to be found in that region'"], e. g. carbonyl-["] and halo-metal complexes" "I. Furthermore, bending vibrations also appear in the FIR region when light atoms (elements of the second and third periods) are involved; as an example may be mentioned the bending vibration by which the COC angle in methyl formate or methyl acetate is altered (325 or 303 cm- I ) . Still lower wave numbers are recorded if whole groups of atoms vibrate completely or approximately in phase with the atoms under consideration; this can be shown e.g. for alkyl formates when the alkyl group is larger than a methyl group[131.In this case the term skeletal vibrations is also used (see Section 2.3). I / I t-- I -----> Fig. I . Schematic representation of the ring puckering vibration of a four-membered ring (for the sake of simpltcity the substituents on the ring atoms have been omitted). In the absence of double bonds the force constant for this type of vibration is determined mainly by the interaction forces between the substituents on the various ring atoms. The magnitude of these forces is of the same order as that of intermolecular interactions; this explains the low wave numbers of ring-puckering vibrations which, for example, are observed for oxetanes between 50 and 160 cm"I'']. Immediately double bonds are present in the ring, the respective vibrations occur at higher wave numbers. 2.5. Torsional Vibrations In torsional vibrations (as in ring-puckering vibrations) the bond lengths and bond angles also remain, to a first approximation, unchanged. What is changed is a dihedral angle (Fig. 2), i.e. the orientation of two parts of the molecule relative A n g m . Chrm. I n r . Ed. Engl. 1 Vol. 15 ( 1 9 7 6 ) No I to one another; here the only degree of freedom is the rotation of the two parts of the molecule (AX2, BY2) around the chemical bond (A-B) connecting them. The wave number t for the torsional vibration depends on the reduced moment of inertia I , and on the torque constant D : If x molecules containing a total of n,=n.x atoms come together to form an association, then N, normal vibrations are observable where N, = 3 n, - 6 Together with expression (3), this gives: N X - x . N = 6 ( x - 1) Here the reduced moment of inertia is obtained from the moments of inertia of the two parts ofthe molecule analogously to the way in which the reduced mass of a diatomic molecule is obtained from the two individual atomic masses. Thus, not only the mass of the atoms taking part in the vibration but also their distance perpendicularly from the rotation axis is responsible for the wave number 7 . Y Fig. 2. Schematic representation of the torsional vibration of a molecule X2ABY, around the axis of the A -B bond. In the case of a torsional vibration around a single bond the torque constant with respect to this bond is determined by the interaction between the substituents on one of the bonded atoms (A) and those on the second atom (B); this means that its value is determined by steric factors. Typical examples are the torsion of methyl groups around C-C, the torsion of OH C-0, or C-N single bonds[’7-2’*271, groups in aliphatic and of NH2 groups in primary amines’”31or hydrazines’*‘! The situation is different for torsional vibrations around a bond axis that has double-bond character. The rotatability around such bonds is more restricted, and the torque constant can be determined largely by the double-bond character. As has been shown by investigations of butadiene and glyoxal and compounds derived therefrom (C-C bond axes with partial double-bond chara ~ t e r ) [ ’ ~ . and ’ ~ ] of alkyl esters of carboxylic acids (C-0 bond axes with partial double-bond ~haracter)’~’~, the wave numbers for the corresponding torsional vibrations depend largely on the properties of the substituents, which can exert an effect through the reduced moment of inertia and the torque constant (inductive and mesomeric effects). 2.6. Intermolecular Vibrations The number N of normal vibrations of a nonlinear molecule composed of n atoms can be calculated from the formula: N=3n-6 A i y e w . Chem. / ! i t . (3) Ed. Etlyl. i Vol. 15 ( 1 9 7 6 ) N o . 1 (4) In the association there are thus 6(x- 1) normal vibrations more than in the total of x isolated molecules. Where d o these 6(x- 1) vibrations come from? The total number of degrees of freedom concerned with rotation, translation, and vibration is constant for a system of n . x atoms independently of how these atoms are combined into molecules and associations. On combination of x molecules to form an association 3 (x- 1) rotational degrees of freedom and 3 (x- 1) translational degrees of freedom disappear. An individual molecule can no longer undergo translation or rotation independently of the others; only the association as a whole has the possibility of this type of movement. These 6(x - 1) lost rotational and translational degrees of freedom have become the above mentioned 6(x - 1) additional vibrational (librational and translational vibrations) degrees offreedom in the association. They correspond to the so-called intermolecular vibrations in which the individual molecules vibrate as more or less rigid structures. In this process relatively large masses, namely those of the molecules, move in opposition to relatively weak forces, namely the intermolecular interactions; and this results in low-frequency vibrations generally found in the region below 300cm-’. The force constants have values between 1.0 and 0.001 mdyne/A and are thus normally smaller by orders of magnitude than the force constants of principal valence bonds (ca. 20 to 1 mdyne/A). As is well known, the influence of intermolecular interactions on single molecules can be studied by various spectroscopic methods such as MIR, UV/VIS, and NMR spectroscopy. All these methods, however, have in common the feature that they supply only part of the Information-they indicate how the skeleton and the electron distribution of a molecule change under the influence of specific intermolecular interactions, but information about the nature of these interactions themselves is obtained only by investigation of the lower-frequency vibrations, i. e. by FIR spectroscopy and Raman spectroscopy. Applications of FIR spectroscopy for such purposes are described in Sections 2.6.1 to 2.6.4. 2.6.1. Hydrogen Bonding Association by means of hydrogen bonding is observed for water, alcohols, carboxylic acids, amines, etc..[*’l and for their with other polar compounds in the liquid and solid phase, and frequently also in the gaseous phase. The force constants for vibrations of the individual molecules in the direction of the H bonding are between 1 and 0.1 mdyne/A; the force constants of the other so-called Hbonding “vibrations” are correspondingly smaller. Study of H bonding by FIR spectroscopy has been particularly detailed for carboxylic acidsL2’ 3”1. Intramolecular H bonding is regarded as a special case of this type of intermolecular interaction: as a kind of intermolecular interaction it is observable 27 in the FIR spectrum only in the form of intramolecular vibrations. Investigations of compounds of the salicylic acid type are still in progress[32! 2.6.2. "Weak" Intermolecular Interactions Dispersive forces and dipole-dipole interactions etc. belong to this group. The force constants are generally smaller by 1 or 2 orders of magnitude than those for H bonding, and the wave numbers of the corresponding vibrations are therefore very low (less than 100cm 34! Under the usual conditions the absorption bands observed are strongly broadened because of the multivarious intermolecular force constants in solid and liquid samples (cf. Fig. 3). Of particular interest in this connection are rc-electron-rich (dispersive forces!) and strongly polar substances (dipole-dipole interaction !). TO exhaust the information content of the FIR spectrum additional experimental work (matrix isolation techniquel8'1) is necessary. 2.6.4. Lattice Forces in Molecular Crystals Molecular crystals can be regarded as associations in which a very large number of molecules are joined together, whereby various types of interaction force can be observed. The intermolecular vibrations of such a crystal are known as lattice vibrations. Only those lattice vibrations in which neighboring primitive unit cells vibrate in phase can be observed by means of IR and Raman spectroscopy. This means that when the number of intermolecular vibrations of a molecular crystal observable in IR or Raman spectra is being determined, only the number of molecules combined in one' primitive unit cell is taken into account; this number replaces the x in Eq. (4). I n this case equation (4) must be modified further. Thus, not only the librational and translational vibrations within a primitive unit cell but also the three in-phase librational vibrations of the whole primitive unit cell around the three spatial axes can be observed in the IR and/or the Raman spectra. This is. ofcourse. not the case for the three in-phase translational vibrations of the whole primitive unit cell in the direction of the three spatial axes; these correspond to translatory movement of the whole crystal. The number of lattice vibrations of a molecular crystal that can be observed by the methods of vibration spectroscopy thus amounts not to 6(x- 1 ) [Eq. (4)]. but to 6s-3. The spectral region corresponding to these lattice vibrations lies below 2(jocm-lL33.3X1 The crystalline portions of polymers provide a special problem in crystal structure; in polyethylene, for example, the crystalline portions give rise to a lattice vibration at 72 cm- 1[3q. 401 hr' If the masses of crystalline components are sufficiently large then, of course, also the lattice vibrations of valence and ionic crystals lie in the FIR region (e. g. alkali metal halides such as KBr etc.). 3. The Importance of FIR Spectroscopy in Chemistry and its Problems I I 500 .m 100 3W t-w i i e I 200 1w 20 number [ m11 Fig. 3. FIR spectra of a ) chloroform and b) benzene (liquid' layer thickness I m m ) and c) of acetonitrile (dissolved in CCI,, c % 10 moI-",<;; layer thickness 1 mm). 2.6.3. Interactions between Adsorbed Molecules and the Adsorbent Investigatlon of the vibrations of adsorbed molecules with respect to the adsorbent enables some conclusions to be made about the nature of the interactions between them"'- 371. Inter aha, H bonding, dispersive forces, and dipole-dipole interactions may play a part here.-Such studies are important for surface chemistry and heterogenous catalysis. As examples may be cited studies on the adsorption of ~ a t e r ' ~ ' . "and ~ of organic molecules[3510; silica gel. Admittedly, the information content of the FIR spectra is usually greatly reduced because of the heterogeneity of the surface concerned and the resulting band broadening. 28 All compounds that absorb in the FIR region exhibit absorption bands with considerably higher absorbance coefficients c in the mid IR (MIR) and/or in the UV/VIS region; e.g. in the MIR region E assumes maximum values of about 2000 I/mol .cm, whereas the corresponding values in the FIR region are smaller by a factor of at least 1/5; naturally, this makes itself apparent, on the one hand by the amount of sample required and on the other-so as to ensure an acceptable signal-to-noise ratio in the spectrum-by increased demands on the apparatus. Even more pronounced are the differences in absorbance coefficients on comparing the FIR with the UV/VIS region. It thus follows that quantitative analyses can be carried out with very much higher sensitivity and-as regards equipment-more favorable detection limits in these shorter wavelength regions. Consequently, FIR spectroscopy is resorted to only in exceptional cases for the purpose of qualitative analysis and identification of unknown substances. Its main area of application is in structural elucidations. Although the content of information in the FIR region is undoubtedly smaller than that in the MIR region for many compounds, in particular organic ones, the importance of the FIR region in chemistry, including organic chemistry, is not to be underestimated. The reason is that the assignment A n y r x . Cheni. Iilr. Ed. Engl. I Vul. I5 ( l Y 7 6 ) No. 1 of an FIR spectrum provides access to important physical and physicochemical information: 100 The structure of many compounds that contain one or more heavy atoms can be determined with the aid of vibration-spectroscopic methods only if the FIR spectrum is taken into consideration. - For normal coordinate analysis the information about the vibrational behavior of molecules must be as complete as possible; the low-frequency region cannot be arbitrarily excluded, for the solution is then no more than approximate. - For determination of thermodynamic functions it is precisely the low-frequency normal vibrations that play a decisive role'"']. - Facts about the potential barriers in interconversion of conformers are closely related to the wave number of torsional and ring-puckering vibrations, which are found exclusively in the FIR region. -- Intermolecular interactions, which are often responsible for the conformation of a compound and decisively determine its physical behavior, can be investigated directly in the FIR region. The problems of assignment in the FIR region reside in the facts that a large number of different types of vibration (see Sections 2.1 to 2.6) can be observed next to one another and that the concept of group frequencies, which so greatly assist the assignment in the MIR region, is applicable only in exceptional cases in the FIR region. The reason for this is that the low-frequency normal vibrations generally involve a relatively large number of atoms with non-negligible amplitude. This is immediately apparent for skeletal, torsional, and intermolecular vibrations; for example, in a torsional vibration around a specific bond axis there is movement of all the atoms in the molecule except those that occur along this axis. In the case of torsional vibrations one can therefore speak of a group frequency only when the reduced moment of inertia is almost identical with the moment of inertia of the group under study referred to the torsion axis; i. e. when this moment of inertia is small in comparison with the corresponding moment of inertia of the other part of the niolecu1e.An additional complication in the interpretation of the FIR spectrum arises through the increasing intensity of "hot bands" with decreasing wavelength (at 500 cm- and room tempel-ature already w. 1'10 of the molecules investigated are in the first excited state!l4l1)andthrough band-splitting which occurs. ir7rrr trfitr. in torsional vibrations and has its origin in special Moreover potential curLes having two or three minima" as in the MIR region- difference bands, combination bands. and overtones can interfere: they usually lead. of course, to weak intensity bands. How varied the information in the FIR region can be, even for purely organic molecules (only light bonded atoms), is shown in Fig. 4, in which the FIR spectra of some alkyl acetic esters have been c ~ l l e c t e dThe ~~~ assignment ~. naturally requires the additional introduction of Raman spectroscopy. Further difficulties arise when attempting to extract the maximum of information from the FIR spectrum in the region around and below 100cm-'. The reasons for this are as follows: 50 - - "Hot bands"ofconsiderab1e intensity occur at room temperature in this region of low-frequency vibrations. 500 k%d a1 LOO -Wave 300 200 100 0 number [cm-'l Fig. 4. FIR spectra of a l k k l esters of acetic acid, CI-I,COOR ( d i s d v e d in C J i , L , c :5 to IOmol-",,: layer thickness 2 m m ) : a ) R=CH.,. h) R = C L H i r c) R = i - C , H - . d ) R=t-C,Hq. Low-frequency intramolecular vibrations are strongly influenced by intermolecular interactions. - The low-frequency vibrations corresponding to the intermolecular interactions in solid and--particularly-liquid phases are usually characterized by a large variety of different force constants. As a result, a single broad absorption band is found below 100 c m - for many substances, particularly in the liquid phase (cf. Fig. 3). The information content of this band can possibly be made accessible to the chemist by the matrix isolation technique["]. This opens up a broad and very interesting field of application for FIR Fourier spectroscopy where, however, because of fundamental difficulties (high radiated power on cooled sample!), the Raman spectrum is no longer expected to have its usual importance as a supplementary source of information. The assignment of an FIR spectrum is only possible when the sample to be examined has been suitably prepared. For the problems arising in this connection and their solution the reader is referred to the literature[63,- 3 - 8 2 1 , -- 4. Principles of the Fourier Method 4.1. The General Problem in Optical Spectroscopy The general problem in optical spectroscopy, irrespective of the spectral region, is to determine the intensity I of electromagnetic radiation as a function of the frequency v , the wave number 0 or the wavelength A. The fact that especially in 29 IR spectroscopy the transmittance T($,=I(?, I,,,,,, or its negative common logarithm, the absorbance, is used. is of secondary importance and arises because a substance-specific datum is called for that must be independent of the apparatus-specific intensity distribution lo,?). For thedetermination ofan intensity distribution the series of waves with different wave numbers contained in a polychromatic light wave must be separated from one another in some wayt42! For this purpose a code is needed in which, in accord with information theory[431, a set of values of some measurable quantity A , is assigned to the set of values of the wave number 0, : Thus the intensity is measured as a function of this measurable quantity A,, and the dependence I,?,,is finally determined on decoding. When choosing the code [Eq. (S)], the properties of electromagnetic radiation must be taken into account. For an unambiguous description of a nonpolarized, monochromatic, electromagnetic wave in a vacuum four quantities are available: Intensity I Wave number 0 Direction of propagation 7f Phase state cp I is one measurable quantity of interest here, and 0 is the other. There are thus two determinable quantities still free that can be used for coding, namely ?and cp. CO,)-(Z,) (54 (0,)-(cp,) (5 b) The problem of coding solves itself if an apparatus is available for use as a light source by means of which monochromatic radiation of variable wave number can be produced. The wave number 0, can then always be related to a specific setting X , of one parameter of the apparatus: The decision for or against one or the other form of coding depends on the following criteria: - Unambiguity : - Accuracy; The apparatus required; - The radiation flux through the measuring equipment. The far infrared lies between the microwave region ( < 10 cm - ')and the mid IR region ( > 500 cm- '). The question is whether the methods of coding customary in these last two spectral regions and of long-proved value are applicable and useful in the FIR region. The microwave region is a domain of nondispersive spectroscopy. Electronically controllable high-frequency tubes are used, e. g. reflex klystrons, which act as sources of monochromatic light. For the medium IR region the so-called conventional dispersive spectroscopy is used. By means of a monochromator each wave train is assigned a specific propagation direction ?, corresponding to its wave number 0,. In practice, we use for this purpose: - the wave-number dependence of the angle of refraction of light in a suitable prism or the wave-number dependence of the angle of diffraction of light by a suitable grating At a specific distance from the dispersive element (prism or grating) one can thus, given suitable imaging optics, measure the intensity I,& corresponding to each direction of propagation by means of an aperture diaphragm and, behind it, a detector. The intensity distribution I,<,, is then obtained by means of the well-known code (5a). Assessed by the previously mentioned criteria, this method of coding is satisfactory except in one respect: the radiation flux from the light source to the detector is considerably reduced. There are two reasons for this reduction: - To a first approximation, the aperture diaphragm in front of the radiation detector allows the passage only of light with a specific propagation direction and thus with a specific wave number Oi. All light waves with 0+Oi are kept away from the detector. Deflection of the light at the dispersive element occurs in one dimension ; accordingly, the diaphragm aperture delimiting the radiation flux muxt be a slit. The finer the details of the intensity distribution I , ; , that are to be recorded, the smaller must be the slit width. The radiation flux is thus reduced in comparison with that from a comparable circular aperture (see Section 4.2). A circular aperture can always be used if the code (5a) is not required, i.e. if the wave trains of differing wave numbers all have the same direction of propagation. So long as both intense light sources and sensitive detectors are available for the spectral region under study, the reduction of the light flux is really disadvantageous (that is the signalto-noise ratio is unpdvorable and,'or the time of the measurement is prolonged) only in exceptional circumstances. These exceptions occur: when there is a strong background absorption ( e . g . by the solvent), - in investigations involving high-resolution conditions, and in the investigation of microsamples. Spectral regions, in which the conventional dispersive method, i. e. code (Sa), leads in most cases to useful results, are the near-IR and the UV/VIS as well as the mid IR. - ~ ~ 4.2. Principle of the Fourier Method Regarding Its Use in FIR Spectroscopy ~ 30 The nondispersive spectroscopy practised in the microwave region is not applicable to the FIR region; electronically controllable high-frequency tubes are available only up to a maximum of a few 100 GHz (100 GHz 5 3.3 cm-'). It can be expected, however, that in a few years controllable lasers in the FIR region will take over the function of electronically controllable high-frequency tubes in the microwave region.-If conventional dispersive spectroscopy is introduced in the FIR region [coding according to relation (Sa)], the reduced radiation flux is disadvantageous even under normal conditions, since only light sources of relatively weak intensity and relatively insensitive radiation detectors are available for the spectral region below 500cm '. For the FIR region we thus need a different kind of codingC4',4h1. This is provided by relation (5 b), in which a specific phase or phase difference is unambiguously assigned to each wave number while, however, all the wave trains of different ~ wave number retain the same direction of propagation. The concept of phase difference makes it immediately clear that there must be two waves between which the phase difference exists. Hence the light wave to be investigated must be divided into two partial waves. Let us first assume-for the sake of simplicity-that a monochromatic plane light wave is present and that it has been divided into two partial waves of the same intensity I,i),same direction of propagation, and same state of polarization. After we have produced a variable phase difference cp between them, we now make these partial waves overlap. The cp-dependent intensity is then measured in the overlap region: If cp=O or an integral multiple of 271, then I(,+) assumes its maximum value of 41(?,. If cp is an odd multiple of n, then has the lowest possible value, namely zero. 11,) is called the interferogram, and the cp-dependent component in formula (6) is called the interferogram function: For the production of the variable phase difference between the two planar light waves the latter are allowed to traverse different but well-defined paths. The phase difference cp depends on theoptical path difference x and on the wavelength h or wave number 0: X cp1.,=2n- = 2 n 0 x h (8) Thus, the desired coding is achieved: according to Eq. (S), the wave number 0 is unambiguously assigned a phase difference cp as a function of the optical path difference x. In the present experiment there is thus, according to Eqs. (7) and (S), for the monochromatic light wave under study, a corresponding experimentally measurable harmonic function-the interferogram function F,,,-whose amplitude is proportional to the desired quantity I , i , and whose argument is equal to the phase difference cpl,,=27c0x: If, in place of monochromatic light, we are dealing with polychromatic light having any desired continuous intensity distribution, the interferogram functions corresponding to the individual wave numbers overlap without distortion with a fixed phase relationship, given by Eq. (S), between them. For a n optical path difference x = O all the wave trains, irrespective of the wave number 0,give an interference maximum. Mathematically speaking, the overlap of the interferogram functions corresponding to the different wave numbers implies integration of the “monochromatic interferogram function” (7 a) over all possible values of 0: Here Ilelhas the meaning of a n intensity density, i. e. exactly the quantity that one wishes to determine in spectroscopic investigations. In Fig. 5 the interferogram functions for polyA t l g r ~ Chwi. Int. E d . Enyl. I_ lie/. i j (1976) No. 1 chromatic light having a continuous intensity distribution and for monochromatic light are placed side by side. It will be seen that in the former case the amplitudes become smaller and smallerwith increasingabsolute valueof the path difference x and finally reach zero; the cause of this is explained in Section 4.3. x:a Optical I ~ path differencex Fig. 5. lnterlerogram functions (sections) for polychromatic light having a continuous intensity distribution (above) and for strictly monochromatic light (below): the ordinate scales are not all the same. After its experimental determination, the “polychromatic interferogram function” [Eq. (9)] must be investigated for its harmonic components (characterized by the wave number 0 ) whose amplitudes all afford the desired value I(cj. The procedure that makes this possible is none other than a mathematical filtering of the function F based on the property of orthogonality of harmonic functions (cf. standard works on higher mathematics): The method outlined here for investigating an intensity distribution I,cjthus differs in fundamentals from conventional dispersive spectroscopy only in that the optical filtering of the polychromatic light wave (in the monochromator) has been replaced by mathematical filtering of the “polychromatic interferogram function” provided by the interference experiment. Independently of whether the filtering is carried out by the optical or the mathematical method, we are always dealing with a Fourier analysis. The method in which the filtering is mathematical is called Fourier spectroscopy or Fourier transformation spectroscopy (FTS), because here, in the determination of an intensity distribution, the procedure of Fourier analysis or.Fourier transformation clearly emerges in its own right.-In both methods interference experiments are carried out (in dispersive spectroscopy with multibeam interference, in Fourier spectroscopy with two-beam interference). In both methods the desired intensity distribution is provided by evaluating the interference pattern. Both methods require Fourier analysis; in conventional dispersive spectroscopy this procedure is effected by the analog monochromator/detector system, in Fourier spectroscopy by a digital or analog computer after or during the recording of the experimental quantities by 31 Parabolic mirrui I Sample position II _---______-_------- Golay detector - ,-Light - source with aperture I I I I diaphragm !I 1 I I ‘ ! I Blackened polyethylene lens I I II 1 /‘ I I I 1 I I I I I . ~ I I 1 I Fixed mirror I I I Oetector unit I Beam splitter .‘ Light source housing Reduction optics Beam splitter unit Sample space Fig. 6. Construction and optical path of a Michelson interferometer for spectroscopy in the FIR region. the detector. The physical and mathematical bases of the Fourier method have been clearly described by, inter a h , Gunzel, Geick, Hurley, Grossu, and H o r l i ~ k [ ~” I ~. .S~pec~ific aspects of Fourier spectroscopy are dealt with in the proceedings of the “Aspen International Conference on Fourier Spectroscopy I 970‘521. What, then, are the advantages of the Fourier method over the conventional dispersive method? - In the first place, all wave trains with their different wave numbers reach the detector simultaneously and not in succession. This difference in comparison with conventional dispersive spectroscopy is called “the multiplex advantage” (“Fellgett advantage”). - Secondly, there is no need for the diaphragm aperture to be in the form of a slit, since all the wave trains with their different wave numbers have the same propagation direction. With a circular aperture a radiation flux at least two orders of magnitude higher than with a slit aperture can be achieved for optical systems having the same resolving power (cf. Section 4.3) and the same collimator parame441. In this connection one ters (focal length, speaks of “throughput advantage” (“Jacquinot advantage”, “etendue advantage”, “aperture advantage”). The throughput advantage is naturally partly or wholly lost as soon as-e. g. in the investigation of microsamples-the preset aperture of the apparatus is reduced. Only through the Fourier method has FIR spectroscopy attained its present importance in chemistry for structural determinations. Besides the above-mentioned advantages over the conventional dispersive method, it should be noted that the problems of optical filtering are more easily solved and especially that the optical systems are comparatively simpler (Fig. 6). In this respect the Fourier method is also superior to Hadamard transformation 46. 31 which otherwise is, or at least could be, of equal value. It remains, however, to explain why the Fourier method is not also generally used for the IR and UV/VIS regions. An important reason for this-among o t h e r ~ [ ~ ~ ] - the is existence of problems with precision, which are not very impor4s3 32 ’ tant between 20 and 1000pm (the FIR region) but become greater with decreasing wavelength; the shorter the wavelength of the light under study, the greater is the precision needed SO as not to exceed a definite relative limit of error in the determination of the abscissa values for F,,, [see Eq. (9)]. It follows that even in the mid IR region (between 2.5 and 20 pm) very much greater technical effort must be expended[54] to obtain the full advantage of the Fourier method over the conventional dispersive method. The important components of a Fourier spectrophotometer are the interferometer, which produces the experimental function or I ( x ) and , the system for evaluating the interferogram data. It is not the purpose of this review to go into details of apparatust6*s6 62, 641 (see also Refs. [2-41) or questions of evaluating the interferogram (Fourier transformation)[s7,65-69. 721;here we shall only relate the theory discussed in this section to practice. Figure 6 shows schematically the construction of an interferometer used for FIR spectroscopy (Beckman IR-720) and the optical path of the radiation. It functions on the Michelson principle. The two partial waves with the properties described above are produced by splitting the wave amplitude of a plane light wave in the beam splitter unit-this is the only component of the interferometer that is important for the explanation of the principle outlined above. (It can naturally also be the wave front that is split, as occurs in the lamellar grating interferometer[60-62!) The beam splitter itself is a film that absorbs as weak]; as possible in the spectral region of interest and which permits part of the incident light to pass through while reflecting the remaining part. The light transmitted through the beam splitter falls on a fixed plane mirror parallel to the wave front. The light reflected from the beam splitter falls on another plane mirror, also parallel to the relevant wave front. Unlike the first mirror, however, the second mirror is movable in the direction normal to the wave front of the beam incident upon it, and its position determines the phasedifference between the beams transmitted and reflected at the beam splitter. After respectively reflection and transmission through the splitter the two partial beams ~ Angrw. Chern. inr. Ed. Engl. j Vol. I S ( 1 9 7 6 ) No. I finally reach a region (to the left of the splitter in Fig. 6) where they can overlap (see Section 4.3). Their intensities are exactly the same, since, given ideal mirrors and optimal adjustment, they have the same history. In the region where the two partial beams overlap the resulting intensity can be measured and related to the position of the movable mirror, i. e. the interferogram I,,, can be measured. Thence the interferogram function F(x, is obtained by simply subtracting a constant value (which is equal to the integral intensity of the two partial beams), and finally the spectrum 1 , ~by ) subsequent mathematical filtration (Fourier transformation) of F(x). All other commercially available FIR Fourier apparatus (Polytec Model FIR-30, Digilab/Cambridge Instruments Models FTS-16 and FTS-18, Bruker-Physik Model IFS-114, Coderg Model FS-4000, and Grubb Parsons Models IS-3 and Mark 11)-in one or other variant-also function on the Michelson principle. Such a grating produces very broad interference maxima. This means that generally speaking a very large number of diffraction maxima must be passed through before the first occasion is encountered where the maximum for one spectral line coincides with the minimum of the other, which is the prerequisite for the fact that the existence of two adjacent spectral lines is at all observed in this experiment. As implied in Eq. (Ila), the grating order n must be the higher, i.e. the optical path difference n.h between the two interfering beams must be larger, the closer together are the two spectral lines, i. e. the smaller is A5. This result can be applied to the Fourier method: in place of n.h we set the maximal optical path difference xmaX between the two light waves that are made to overlap: 4.3. The Resolving Power of the Fourier Method The resolving power A obtainable by means of a grating can be calculated from the expression : 0 In order that two infinitely sharp spectral lines separated by A 5 can just be resolved, x , must have the value prescribed by Eq. ( l l c ) : A=-=nN A0 where A 5 is the distance [measured in cm-’) between two infinitely sharp neighboring spectral lines that are just resolved, 5 is the mean wave number for these two lines, N is the number of lines of the grating, and n is the grating order. xm.x 1 2A0 = __ A0 can of course also be regarded as the experimental broadening, depending on the value of x,,,, of an infinitely From l i g h t source Lattice with 2 Lines Detect Or - ‘ I Beam splitter I Movable mirror ; I I ,1 A\ _ _ , 2-, __ 11x1 h f i x e d mirror Fig. 7. Schematic representation of the phenomenon of two-beam interlerence on a grating with two lines (above) and in the Michelson interferometer (below). On the right is sketched the interferogram produced in both cases for strictly monochromatic light In the Fourier method a two-beam experiment (cf. Fig. 7: beams 1 and 2) is carried out. So far as the resolving power is concerned this amounts to a diffraction experiment with a grating (Fig. 7) containing only two lines (N=2); in this case we have: A I I ~ Y I I Cheiii. . l n r . Ed. Engl. f Vol. 15 ( 1 9 7 6 ) No. I sharp spectral line. This is the quantity that in conventional dispersive spectroscopy is usually given as the “spectral slit width”. In practice the spectral lines are never infinitely sharp, a fact that is connected with the finite breadth of the energy levels between which the light emission processes take place. 33 The question arises how the finite breadth of the spectral lines is expressed in the interferogram function. For this purpose the Heisenberg uncertainty principle provides an answer: where Ap is the uncertainty in momentum and Aq the uncertainty in the position of a particle (e.g. a photon) and h is Planck’s constant. Taking into account that where AF,,, represents the half-width of the emission lines studied, it follows from relation (12) that: to exactly half of the maximal possible interferogram value (at x = 0). The continuous decrease in the interferogram function, which originates in the limited coherence length of the light wave, is called the “seIf-apodization”[’ ‘ I . So far only more or less sharp emission lines have been discussed. In vibrational spectroscopy, however, one has to do mostly with continuous emissions weakened by absorption in specific areas. What, then, is the value of the coherence length Ifor such an intensity distribution, absorption occurring only within a specific spectral interval of width 2A\O,,, (Fig. 9, top) and there has the form prescribed by the transmittance IT;?,? The coherence length can be determined by carrying out the interference experiment described in Section 4.2 and thus measuring the region in which the interferogram function depends on the optical path difference. t For photons we can replace A4 by the length of the wave train under consideration, namely the coherence length 1. It follows that The coherence length of a light wave is the greater, the smaller the half-width of the sharpest of the spectral lines corresponding to the light wave. m x- 0 : ;J,+ A? ”2 So- A?,,? Fig. 9. Determination of the coherence length of light corresponding to a continuous intensity distribution weakened in a narrow spectral interval Z A G , , (for explanation see text). The interferogram function for the intensity distribution discussed here has the following mathematical form [see also Fig. 9 and Eq. (9)]: x = I’ 2’ La_q2al Fig. 8. Schematic representation of the overlap region between two wave trains of coherence length I in dependence on the optical path difference Y. The two partial waves produced from the light wave to be studied (cf. Section 4.2) can only interfere if they overlap completely or partially. Complete overlap occurs (see Fig. 8) when the optical path difference x is zero. The larger is x, the smaller the overlap region becomes and the less marked are the interference phenomena. Accordingly, the modulation amplitude of the interferogram function decreases continually with increasing optical path difference (Fig. 5, upper part), until it becomes zero as soon as the optical path difference equals or exceeds the coherence length (x 2 1). In the region where x> I, the interferogram assumes a constant value which is independent of the optical path difference and corresponds 34 The subscript k refers to the unweakened continuous intensity distribution. This equation can be rearranged to: Angrw. Chem. I I ~ IEd. . Enyl. , &I/. 15 (1976) N o . 1 Fix1 cI with Fig 10. Determination of the sampling intervals (for explanation see t e x t ) and In order to be able to calculate Zlv) from F,:, by Eq. (10) one must care for numerical acquisition of FIX,between x=O and x=xmax,which is not possible continuously but only at specific intervals Ax (Fig. 10). This procedure in which Eq. (15) is used with m=O, 1,2... we obtain (1 5 ) x=m.Ax The interferogram function F,,, thus has two additive components : - the contribution Fk(xl, which corresponds to the broad-band continuous intensity distribution (Fig. 9, upper part) and which decreases “rapidly” to zero (Fig. 5, upper part), and - the contribution FA^] 21n), which, except for the minus sign, corresponds to an emission band with the intensity distribution z k ( v ) ( 1 - 7&,) and half-width AF1.2 (Fig. 9, lower part) and which decreases to zero the more “slowly”, the smaller is A? ,2 (half-width of the absorption band). The coherence length 1 of a light wave that corresponds to a continuous intensity distribution weakened in the interval 2AF1‘2 by absorption (Fig. 9, upper part) can thus be estimated [Eq. (13)] if an indication of the half-width of the sharpest absorption bands to be expected is available. For solid, liquid, or dissolved samples under normal conditions, experience shows that the half-width achieved is less than 10cm-I only in exceptional cases. With this standard factor Eq. (13) leads to a value of ca. 1 mm for the coherence length. If the maximal optical path difference is chosen as xmax=2mm,then the total region O < x < l in which the interferogram function depends on the optical path difference is usually covered. If xmax< [, then, as when the spectral slit width in conventional dispersive apparatus is made too large, experimental broadening of the absorption lines is observed, as well as disturbing artefacts-though the latter can be reduced mathematically (by apodization)“’, ’‘I. If xma.> 1, then the measurement time is increased without any gain in spectral information, and in the region I<x<xrnaxadditional noise is recorded which contributes through the Fourier transformation to a worsening of the signal-to-noise ratio throughout the entire spectrum. is called digitization of the function F,x,, Ax the sampling Due to the digitization of the interferogram function it is, of course, necessary to convert the integral in Eq. (10) into a sum. After the choice of Ax’ the upper limit of the spectral region to be measured is already fixed because, according to the sampling theorem of information theory, harmonic functions can be characterized unambiguously in digital form only when at least one ordinate value per half-period is sampled. The Fourier transformation of F ( x ican thus be effected unambiguously only if this condition is fulfilled for all harmonic functions making up F,,). Since, however, the periods of these harmonic functions are exactly equal to the wavelengths of the polychromatic light studied, only light of wavelength h>h,,, may contribute to the measured signal Z,Kl; we have: or Wave trains with F>Fmax contribute intensity to the wave trains with F < F,,,; this means that the form of coding [relation ( 5b)] is not unambiguous. Unambiguity can, however, be brought about if all the intensity contributions with wave numbers 0>Om,, are eliminated by optical or electrical filtering. The contributions with wave numbers Fz>F,,, to the intensity expected at wave numbers 0 <,F ,, can be determined as follows: The “monochromatic interferogram function” F Z cos 2 T ~ S1,x is assigned to the wave number F1. From this, with allowance for Eqs. (15) and (16), we obtain 4.4. Spectral Region for Measurement by the Fourier Method mnFl F1,,,= zcos 7 The spectral region usable for measurements is determined, as in every spectroscopic method, by the characteristics of the radiation from the light source, the sensitivity characteristics of the detector, and the type and efficiency of the optical apparatus. A further aspect, however, also plays an important part in the Fourier method. Aagru.. Cheiii. I n t , E d EngI. ! Vol. 15 ( 1 9 7 6 ) N o . I Vmax Then we determine all those “monochromatic interferogram functions” vmnx 35 that with F2>Fmaxhave the same value as Fl,mlat the same m ; after digitization they can no longer be distinguished from F1(,,,).Finally, the calculation gives: 0, ={ In a search for errors not only the spectrum but also the interferogram should be thoroughly and precisely inspected. Differentiation of the various types of error (e. g. electronic disturbances, wrongly chosen instrumental parameters, unsuitable preparation of the sample) is often only possible by checking the interferogram alone (“Qualitative Interferogram Evaluation”)[82’. + (2n 2)Fm,,- F 1 2n 0,,,+0, If it is not certain that all the wave trains with F>Dmax are excluded by optical or electrical filtering, then the intensity measured for O,<Fmax may arise wholly or partially from intensity contributions with 2 P m , , ~ 0 1 40,,,+01, , ... efc., the so-called false light contributions. The problem of false light contributions arises not only for the Fourier method but also in conventional dispersive spectroscopy. Here the unambiguity of the assignment (5a) can be safeguarded only by optical filtering, which is most efficiently achieved with double or triple monochromators. The false light (also misleadingly called stray light) that has to be filtered out arises mainly from interfering grating orders. Example I Figure 11 shows that a slightly structured, continuous intensity distribution I(?) provides an interferogram function F ( x )that is dependent on the optical path difference .x over only a relatively narrow region (Fig. I l a ) . On the other hand, the optical path difference at which the interferogram function becomes constant (e.g. identical to zero) is the larger, the sharper are the spectral lines present in the intensity distribution I,,, (Fig. 11b, rotational transitions of water vapor). Interterogram Spectrum 500 rn 1 I 1 400 300 200 t_ 100 I 20 -0 1 0 I I I 0.1 02 0.3 Optical path diflerence [cm-’] W a v e number [cm-’l --is Fig. 11. Comparison of the interferogram and the spectrum in the cases o l ( a ) continuous. and (b) discrete spectral intensity distributions. A further contribution is provided by light that passes through the monochromator by routes deviating from the prescribed beam path. 5. Simple Examples of the Interpretation of Interferograms T o obtain a spectrum on a Fourier spectrophotometer the Fourier transformation must in principle be carried out quantitatively. This is possible only with the aid of data-processing machines, except for strictly monochromatic radiation (Fig. 5, lower part) or for polychromatic radiation that contains only a few, strictly monochromatic, components, neither of which are of practical interest. After some practice, however valuable qualitative information can be extracted from the interferogram without a quantitative Fourier transformation. The labor of qualitatively interpreting interferograms will materially assist the beginner in appreciating the principle of the Fourier method. 36 The explanation of this follows from the Heisenberg uncertainty principle (Section 4.3), which provides a connection between the coherence length of a light wave and the half-width of the corresponding spectral lines or bands. Thus, one can infer from the interferogram function alone whether there are more or less sharp emission or absorption areas in the intensity distribution studied. Example 2 The periods of the harmonic functions that form the interferogram function are the larger, the longer are the wavelengths of the light studied. This can be seen in Figure 12. The figure shows two intensity distributions-one mainly short-wave with a mean wavelength of about 26 pm (386 cm- ’) and a second, mainly long-wave, with a mean wavelength of about 155 pm (65 cm- ‘)-together with the corresponding interferogram functions. The ratio of the mean wavelengths is about I:6 and the same ratio is found again in the separations of the two central minima in the corresponding interferoAngen. Chern. Int. E d . Engl. J Vol. 15 ( 1 9 7 6 ) N o . I Spectrum - Wavenumber [cm-’l lnterierogram Optical path diflerence [cm-’] + Fig. 12. Comparison of the interferogram and the spectrum for mainly shortwave (a) and mainly long-wave light (b) gram functions. The other minima in the interferogram function corresponding to the shorter-wave radiation likewise follow one another more closely than in those corresponding to the longer-wave radiation. Thus, from the interferogram function alone one can decide whether mainly short-wave or mainly long-wave light is concerned. In this way one can e. g. determine whether the attempts t o exclude light-scattering effects during the preparation of the sample were successful. Example 3 Periodic interference of the FIR spectrum prejudicial to the substance-specific character of the information obtained is often observed; this can have various causes, which can be unambiguously identified from the interferogram function: - Disturbing impulses in the interferogram function (Fig. 13a); they are caused by electrical or mechanical interferences and are not symmetrical about the position x=O. - Internal reflection in a plane-parallel layer (e.g. in the cell, in the sample carrier, or in the sample itself18’.821);figure 13b illustrates this case: the spectral intensity distribution corresponds to the apparatus characteristic of a Michelson interferometer where a plane-parallel almost absorption-free paraffin wax film about 100pm thick has been introduced into the path of the beam. The internal reflection in a single plane-parallel layer produces in the interferogram two side-burst peaks arranged symmetrically around x = O (arrows in Fig. 13b). They are converted by the Fourier transformation into the periodic interference observed in the spectrum (“channel spectra”). If we use for the Fourier transformation only that part of the interferogram function that lies between these two maxima, the interference in the spectrum disappears; admittedly, in certain circumstances the resolving power is then too small for the determination of the true band form (see Section 4.3 : cf. Fig. 1 3 b and 13c, absorption band at 72cm- ’). The plane-parallelism between the relevant surfaces must then be destroyed in some other way; in the present example this was done with the Angen,. CArm. I n t . E d Engl. j Vol. 15 ( 1 9 7 6 ) No. I KTZD c Wave number [ c m - l l Fig. 13. Causes of periodic interference of the FIR Fourier spectrum (for explanation see text). 37 - paraffin wax film (wedge shaped), which finally gave the interferogram and the spectrum shown in Figure 13d. Overlapping ofa boxcar function (error in the determination of the interferogram function from the interferogram) or a triangular function (so-called drift in the interferogram) on the interferogram function used for the Fourier transformation[' 21. 6. Concluding Remarks Introduction of the Fourier method into FIR spectroscopy has not only led to a considerable facilitation of work and an improvement in the quality of the spectra, but has also fostered interest in the FIR region. This is the case especially for the introduction of FIR spectroscopy into organic chemistry. Owing to the absence of heavy bonded atoms in organic molecules, the information content of the FIR spectrum was long underestimated. It followed that the demands on finance and time made by the measurement of FIR spectra by the conventional dispersive method were rightly assessed as too great. The introduction of the Fourier apparatus has certainly not eased the financial burden, or at the most slightly so in isolated cases; but its efficiency-greater than that of the conventional dispersive apparatus-has encouraged attack on, and indeed in some cases the solution of, the problems set out in Sections 2 and 3. The region below 100cm-', which is especially important for the study of weak intermolecular interactions, would certainly have remained inaccessible had it not been for the Fourier apparatus, and the expansion of collections of spectra far into the FIR regionLs4]would presumably never have happened without the Fourier apparatus. In recent years the Fourier method has received recognition and more extended use in NMR spectroscopy[65.8 5 - 8 8 1, as it has in FIR spectroscopy. In particular, 3C-NMR spectroscopy owes its present importance as an analytical tool exclusively to the Fourier method. This progress report is based on five years experience in thefield o f F I R Fourier spectroscopy which the author collected in the Anwendungstechnik and Training Department at Beckman Instruments GmbH in Miinchen (Germany). Opportunity is taken here to express thankful recognition of the possibilities ofered by the firm Beckman to carry out relevant scientific work-particularly in the field of molecular spectroscopy. Thanksaredue to Prof. Dr. B. Schrader (Universitat Dortmund) and Priv. Doz. Dr. P. Bleckmann (Universitat Dortmund) for valuable suggestions and e.\-pert opinion on parts of' the Iwnuscript, and to Grad. Ing. K . Dabelstein (Beckmann Instruments) for help with the measurements necessary for this work. Received: August 4, 1975 [A 92 IE] [I] [2] 13) [4] [S] 38 B . Schrader, Angew. Chem. 85, 925 (1973); Angew. Chem. Int. Ed. Engl. 12, 884 (1973). K . D. Moller and W G. Rorhschild: Far-Infrared Spectroscopy. WileyInterscience, New York 1971. A. Finch, P. N . Gores, K . Radclige, F . N . Dickson, and F . F. Benrley. Chemical Applications of Far Infrared Spectroscopy. Academic Press, New York 1970. R . J . Bell Introductory Fourier Transform Spectroscopy. Academic Press, New York 1973. F . Winflier, Z . Naturforsch. 28a, 1179 (1973). ' [6] R. B. Sanderson and H. E . Scort, Appl. Opt. 10, 1097 (1971). [7] A. R . H . Coirand F. R. Hoiiry, Appl. Opt. 10. 1581 (1971). [8] H . A. Sraub' Einfiihrung in die theoretische organische Chemie Verlag Chemie, Weinheim 1970, p. 194. [Y] J . F.Ferraro: Low-Frequency VibrationsofInorganicand Coordination Compounds. Plenum Press. London 1971. [lo] R. S. McDowcll and L. H. Jones, J . Chem. Phys. 36, 3321 (1962). [ I l l A. Saharini u. L. Sacconi, J. Am. Chem. SOC.86, 17 (1964). [I21 P. R. Brookrs and B. L. Shun., J. Chem. SOC. Dalton Trans. 1972, 1702. [I31 E. Knozinger, unpublished. 1141 K . W F . Kohlrausch. Raman-Spektren. Akademische Verlagsgesellschaft, Leipzig 1943; reprint Heyden and Son Ltd., London 1972. [Is] R . F. Srhaufelr and 7: Shimairourhi. J. Chem. Phys. 47. 3605 (1967). [16] S. I Chan, 7: R. Borgrrs, J . W Russel, H. L. Sirauss, and W D. Girfnn, J. Chem. Phys. 44. 1103 (1966). [17] W G. Fareley and F . A. M i / / e r , Spectrochim. Acta 17, 857 (1961). [I81 W G. F a r r l q and F . A. M i / / e r , Spectrochim. Acta 18. 977 (1962). 1191 W G. Fardry and F-. A . Miller, Spectrochim. Acta 19, 611 (1963). 1201 K . D. M d l r r . A. R. De Meo, D. R. Smrrh, and L. H. London, J . Chem. Phys. 47, 2609 11 967). [2i] W H . Moore,J. R . C . Ching, A. I/: R. Worrier,and S. Krirnm, Spectrochim. Acta 2 9 A , 1847 (1973). [22] S. M. C r a r m and F . F . Benfley, Appl. Spectrosc. 26. 242 (1972). [23] S. M. Crawti and F . F . Benrley, Appl. Spectrosc. 26, 449 (1972). 1241 S. M. Craven. f. F . Benfley, and D . F . Pensenstadler, Appl. Spectrosc. 26, 646 (1972). 1251 W G. Farelc,!, R. K . Harris, F . A. M i l k r , and R. E. Wirkoivski, Spectrochim. Acta 21, 231 (1965). 1261 R . K . Harris and R. E. Witkowsk;, Spectrochim. Acta 20, 1651 (1964). [27] E . Knikinyer, Ber. Bunsenges. Phys. Chem. 78, 1199 (1974). [28] R. J . Jakohsen, J . W Brasch, and Y Mikuwa, Appl. Spectrosc. 22, 641 (1968). 129) G. L. Carlson, R. E. Wirkowski, and W G. Farelrv, Spectrochim. Acta 22, 1117(1966j. [30] R. J . Jakobseii. Y Mikana, and J . W Brasch, Spectrochim. Acta 2 3 A , 2199 (1967); R. J . Jakohsen and J . E. Karotr. ibid. 2 9 A , 1953 (1973). [31] A . Hull and J . L. Wood, Spectrochim. Acta 2 8 A , 2331 (1972). 1321 E. K n B z i n g u , unpublished. [33] B. Schrader, Ber. Bnnsenges. Phys. Chem. ? X , I187 (1974). [34] R. J . Jakohsen and J . W Brasch, J. Am. Chem. SOC.86, 3571 (1964). 1351 I . A. Brodskij, A. E. Sram~cick.and N . G. Yuroslurkij, Rnss. J. Phys. Chem. 44, 138 (1973). 1361 I . A. Brorbkij and A . E Sforrrcirh. Kinet. Katal. 14, 463 (1973). [37] I/: Lorertzelh and G. B. Bonino, Lincei-Rend. Sci. Fis. Mat. Nat. 3X, 312 (1965). [38] A. Anderson and H. A . Gehhie, Spectrochim. Acta 2 1 , 883 (1965). [39] M. 7asumi and S. Krimm, J. Chem. Phys. 46, 755 (1967). 1401 G. J . Saffod and A. 14! Nuurnom. Macrornol. Rev. 2, 1 (1967). [41] G. Herzhrrg' Molecular Spectra and Molecular Structure, Vol. 11. Van Nostrand Reinhold, New York 1945. [42] E. Kniizinger, C Z Chem. Tech. 3, 227 (1974). [43] J . 7: Clerc, R. Kaiser. J . Rend/. H. Spirzy, H. Zrrrier, G. Gorrschalk, H. Mul~ssa,E. Srhwarz-Bergkampf, and R. D. Werdrr, Z. Anal. Chem. 272, 1 (1974). 1441 L. G m x 4 , Z. Anal. Chem. 273. 391 (1975). [45] G. A . van us.^: Multiplex Technlques in Spectroscopy. AFCRL-72-0412, July 19, 1972. Phys. Sci. Res. Pap. No. 502. Air Force Cambridge Research Laboratories, Bedford, Mass. (USA). [46] G. W wing^ J. Chem. Educ. 49, A377 (1972). 1471 L. Genzrl Fourier-Spektroskopie. Plenarvortrage der 33. Physikertagung in Karlsruhe. Teubner-Verlag, Stuttgart 1968, p. 128. 1481 R. G r i d , Fortschr. Chem. Forsch. 58, 73 (1975). [49] W H. Hurler, J . Chem. Educ. 43, 236 (1966). [SO] P. Grosse. Beckman Rcp. 1/70. 3 (1970). [51] G. Horlrck, Appl. Spectrosc. 22. 617 (1968). [52] G. A. Vanusse, A. T: Stair Jr., and D. J . Baker' Aspen Int. Conf. Fourier Spectrosc. AFCRL-71-0019, Jan. 5, 1971, Spec. Rep No. 114. [53] J . A . Decker J r . , Anal. Chem. 44, 127A (1972). [54] P. R . Grifprhs, Anal. Chem. 46, 645A (1974). [55] G. Horlick and H. 1.: Mahnsrurlr, Anal. Chem. 42, 1361 (1970). [56] K . Sakai, Appl. Opt. / I , 2894 (1972). [57] R. C. Milward, Polytec Tech. Bull. 1 11972). [SS] R. C . Milward, Z . Anal. Chem. 264, 195 (1973). [SY] E . Kniizinqer. Beckman Rep. 1+2/74, 23 (1974). [60] R. C. Milword, Infrared Phys. 9. 59 (1969). [61] R. 7: Hall, D. Vrabec, and J . M . Doirling, Appl. Opt. 5. 1147 (1966). [62] J . Strong and G. A. Vaiiasse, J. Opt. Sac. Am. 50, 113 (1960). -[63] M. F . Kinimir : Far-Infrared Techniques. Pion, London 1970. [64] M. J . D. Low. J. Chem. Educ. 47, A163, A255, A349, A415 (1970). [65] D.Ziessow: On-line Rechner in der Chemie-Grundlagen und Anwendungen in der Fourierspektroskopie. Walter de Gruyter, Berlin 1973. [66] P. R . Rorh. Hewlett Packard J . 21. 2 (1970). [67] G. Giinther. Elektron. Datenverarbeitung 6, 275 (1969). ' Angriu. Cheni I n t . E d . Enyl. j Vol. 15 ( 1 9 7 6 1 N o . 1 [hX] [69] 1701 [7 I J [72] [73] [74] [75] [76] [77] [7X] J M. Coolrj and J . W! f i ~ k e r Math. , Comput. 19. 297 (1965). 4t. L. Foriiioii. 1. Opt. Soc. Am. 56. 978 (1966). C. H o r / i c k . Anal. Chem. 43. 61A (1971). E . C. Codiliriy and G. Horlick. Appl. Spectrosc. -77. 8 5 (1973). J E. Hnffiiiariii Jv. Appl. Opt. 8, 323 (1969). If. J . Slounrand K E . Sririr, Beckman Applications Data Sheet, IR-8084. E. Kiiikirzger, Chem. Ztg. 95. 983 (1971). H. R. W i . ~ s ,R. D. MIivrler. and H. H . Cii~irlmrd,Spectrochim. Acta 20. 573 ( 1974). C. Sdiirle, Beckman Rep. 1.'65, 6 11965). D. F . Perirrii.~rur/ln.C. L. Corlson, M! G. Fordo., and F. F. Bmrit'i. Spectrochim. Acta X A , 183 (1972). .M. E PetuAiri, Appl. Spectrosc. 25. 502 (1971). 1791 C. S c h i ~ l eand K . Hulfur. Appl. Spectrosc. 19. I63 (1965). [SO] E . K i i i i ~ m g e rpublication . being prepared. [81] 6. C . H o i ~ i i ~ o r Appl. d, Spectrosc. 23. 620 (1969). [SZ] E. KiiB;iiiqrr, Z. Anal. Chem., in press. [X3] H . E. Hullurn' Vibrational Spectroscopy of Trapped Species. Wile). London 1973. 1841 B. Sdirudo. and W Merrr RamanilR Atlas Verlag Chemie, Weinheim 1974. [ 8 5 ] H . Giinthw, Chem. Unserer Zeit 8. 45 (19741. [86] E. Breitmoirr, G. J i i r i g , and M! L'bulrrr, Angew. Chem. 83, 659 (1971). Angew. Chem. Int. Ed. Engl. / I ) . 673 (1971). [X7] D. A . Nrtzrl. Appl. Spectrosc. 26, 430 (1972). 1881 T C. Foi.ror. Anal. Chem. 42, 109.4 (1970). C 0 M M U N I CAT1 0 N S Treatment of P,N4F,(NHJ2 ( 1 ) with SOCl, afforded P,N,F,(NSO), (2). In the presence of pyridine, (2) eliminates sulfur dioxide to give a low yield of 1,5,7,7,1O,lO-hexafluoro3h4-thia-2,4,6,8,9,1 I-hexaaza-I h5,5h5,7h5,1OA'-tetraphosphabicycle[ 3.3.3]undeca-2,3,5.7.9(l),lO-hexaene ( 3 ) . Compound ( 3 ) is a yellowish crystalline solid which sublimes at 25"C/0.5 torr and melts at 57°C. Its composition is proved by complete elemental analysis, and its monomeric structure in the liquid and gaseous state follows from the molecular weight (exp. 350, cryoscopic in C,H,) and the mass spectrum [ M + : m/e 354 (48"/,)]. In the '"F-NMR spectrum two multipletsappear in the intensity ratio of 1 :2, their centers of gravity lying at SF= 59.8 ppm and 66.2 ppm. The spectrum is independent of temperature, an argument strongly supporting the proposed structure ( 3 ) . Procedure. A Cyclophosphazene Spanned by a Sulfur Diimido Group[**] By Herbert M.: Roeskj and Enno JanssenF*] Although the chemistry of fluorinated cyclophosphazenes has been actively studied in recent years"], derivatives having inorganic ring bridges have nevertheless remained unknown. Consideration of a model of P,N4FR revealed that a central bridge of this kind would have to have at least three members. P1N,F6(NSO),['1 (4.2g, 0.01 mol) is heated in the presence of a catalytic amount of pyridine (2 drops) on a water bath for 7 d. Sublimation of the reaction product affords ( 3 ) (0.28 g, 8 "/,). This compound is also formed in low yield during the preparation of P,N,F,(NSCI,), by heating of P+,N,F,(NHz)z and sulfinyl chloride with elimination of SO, and HCI. Received: August 21, 1975 [Z 323 IE] German version: Angew. Chem. 88. 24 (1976) - [I] [2] f 1. Haiiluic The Chemistry of Inorganic Ring Systems. Wiley. New York 1970; S. Punrel and M Beckr-Gorhvirig: Sechs- und achtgliedrige Ringsysteme In der Phosphor-Stickstoff-Chemie. Sprineer. Berlin 1969. H . M! Rouslj. and E . Junasrti, to be published. 2 S0Cll - 4 HCI Tellurium(1v) Iodide: Tetrameric Te41 Molecules in the Solid[*'] By Volker Pauiat and Bernt Krebs[*l The structures of chalcogen(rv) halides investigated up to now are determined largely by the stereochemical activity of the S'", Se'", and Telv nonbonding electron pairs. Gaseous SF, and SeF, (also TeCI, in the gas phase)"] have the monomeric C,, structure predicted by the VSEPR theory"]; in solid TeF,, on the other hand, square-pyramidal TeF, groups are linked uin corners to form chains, the localized lone pair (E) completing octahedral TeF5E c~ordination'~]. The structures of solid TeCI,'"J, TeBr,"], and SeC1,[61contain tetrameric - -. ~ .~ [*] Prof'. Dr. H. W. Roesky and Di-. E. Janssen [*] Prof. Dr. B. Krebs and DipLChem. V. Paulat Anorganisch-chemisches Institut I der Universitit 6 Fi-ankfurt am Main SO, Niederurseler Hang (Germany) [**I This work was supported by the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie. Fakultit fur Chemie der Universitit 48 Bielefeld, Universititsstrasse (Germany) [**I This work was supported by the Land Nordrhein-Westfalen and by the Fonds der Chemischen Industrie. 39

1/--страниц