Absorption Spectrum of Chlorine Dioxide in the Vacuum Ultra-violet BY C. M. HUMPHRIES, A. D. WALSH AND P. A. WARSOP Chemistry Dept., Queen's College, Dundee Received 15th January, 1963 The far ultra-violet absorption spectrum of C102, previously thought to be continuous, has been found to contain three band systems, at 1829, 1628 and 1568 A. Reasons are given why these new systems are ascribed to C102 itself and not, e.g., to a photolysis product. Each of the systems represents a Rydberg transition. The 1829 and 1628 8, systems are associated with the first ioniz- ation potential, the 1568 8, system with the second ionization potential. The upper states of the first two systems (and the ground state of the ClOi ion) have shorter C1-0 bond lengths than the ground state of C102, but OClO angles not greatly changed.The upper state of the third transition (and the first excited state of the ClO; ion) have OClO angles considerably increased from the C102 ground state value, but CI-0 lengths probably not greatly changed. The symmetrical stretching frequency v l is increased in the 1829 and 1628 A transitions from its ground state value; the bending frequency v2 is increased in the 1628 and 1568 8, transitions from its ground state value. The 1568 8, transition appears as a progression of bands with alternate long, short spacing; Fermi resonance between v; and 2 4 may account for this. The absorption spectrum of the C102 molecule in the near ultra-violet has been extensively studied and discussed by Coon and co-workers.l-3 The spectrum in the vacuum ultra-violet has previously been thought to be entirely continuous,4 but we have now discovered three new electronic systems with vibrational structure. EXPERIMENTAL Chlorine dioxide was prepared by heating moist potassium chlorate with oxalic acid, according to directions given by Brays and by Coon.1 After condensing the gaseous products in a trap at -8O"C, C02 was pumped off and the C102 purified by distillation in vacuum.The absorption spectrum was obtained with both a 1 m vacuum spectrograph giving a path length of 2 ni and a dispersion of CCL. 16 &mm and with a 2 ni vacuum spectrograph giving a path length of 4 m and a dispersion of ca. 8.7 A/mm. RESULTS The absorption spectrum has been photographed from ca. 3300 to 1lOOA. At the lowest pressures used, continuous absorption is evident at the shortest wave- lengths.As the pressure is increased, this continuous absorption gradually spreads to long wavelengths until, at the highest pressure used, there is a short wavelength cut-off at about 2000A. At intermediate pressures, three sets of bands are visible. Over the same range of pressure, the bands of the near ultra-violet system become strong on our plates and eventually, at the highest pressure used, spread (from a maximum at ca. 3300A) to about 2700A. The two shorter wavelength sets of far ultra-violet bands have about the same appearance pressure, this being slightly less than that for the longest wavelength set and comparable with that for the strongest near ultra-violet bands. The bands of each set increase in intensity with pressure, though apparently more slowly than do the bands of the near ultra-violet 137138 SPECTRUM OF CHLORINE DIOXIDE system.With grosser increments of pressure than used here, because of the con- tinuous absorption spreading to long wavelengths with increasing pressure it would be easy to miss the far ultra-violet band systems. The first system begins at 1829 A and consists of three bands whose intensity falls off rapidly from the first to the second to the third. These bands (A, B and C) A B FIG. 1.-(a) The 1829 8, bands; (b) the 1628 and 1568 8, systems. The intensity scale is not the same for (a) and (b). occur at 54,689, 55,709 and ca. 56,720 cm-1 respectively, the separations being 1020 and ca. 1011 cm-1. The system is fragmentary because of the decreasing intensity of the bands towards shorter wavelengths and because of the cut-off due to the continuous absorption.A densitometer tracing of the bands is shown in fig. 1.C. M. HUMPHRIES, A. D. WALSH A N D P . A . WARSOP 139 The second band system begins at 1628 A. The bands (D E F G H J K') are shown in the densitometer trace reproduced in fig. 1 and their measurements are given in table 1. Two frequency separations are involved. The main bands (D F H K') form a progression with spacings of 1051, 1091 and 1068 cm-1. Associ- ated with each main band lies a weaker band ca. 520 cm-1 to the violet. Again the first band is the strongest and the intensity falls off towards shorter wavelengths where the system is overlapped by the third set of bands.TABLE FREQUENCIES OF THE BANDS IN THE 1628 A SYSTEM label cm-1 D 61y430-> 521 \>,,,, E 61,951- F G 63,007 - H 63,572--- 62,48 1 ---- / > 526 '\ J ca. 64,087 K' 64,640 The third band system begins at 1568 A. Its bands (I K L M N 0 P Q) are shown in the densitometer trace reproduced in fig. 1 and their measurements are given in table 2. The bands are somewhat sharper than in the other two systems. TABLE 2.-FREQUENCIES OF THE BANDS IN THE 1568 8, SYSTEM label I K L M N 0 P Q cm-1 (4 (b) 63,774 - 64,282 64,778 65,281 65,770 - 66,28 1 66,767 ca. 67,280 - The apparent intensity of band K may be lowered by the presence of an emission line on its short wavelength side ; if so, the intensity of the bands may vary smoothly with wavelength and the bands might be regarded as forming a single progression in CU.500cm-1. Arranged as a single progression, the band separations are not constant but alternate long, short, long . . . (see column (a) of table 2). Alter- natively, the bands might be arranged as a progression I, L, N, P, each with an associated band (K M 0 Q respectively; see column (b) of table 2). The main1 40 SPECTRUM OF CHLORINE DIOXIDE progression spacing is then ca. 1000 cm-1 and the subsidiary spacing cd. 500 cm-1. If this arrangement is adopted, it should be noted that bands K M 0 Q are only slightly weaker than bands 1, L, N, P respectively, and therefore the arrangement presumably conceals progressions in ca. 500 cm-1. In other words, whichever arrangement is adopted, a progression in cu.500cm-1 is present. With either arrangement, maximum intensity is reached at band L. DISCUSSION For the following reasons the three new band systems observed are believed to be due to C102 and not to an impurity or product of photolysis. (i) The systems do not correspond to any of the known band systems of C12, 02, ClO, HC1, CO, C02 and H2O. The absorption spectra of all but the third of these have been well explored in the relevant wavelength region. (ii) The second band system involves two upper-state vibrational frequencies and must therefore be due to an absorbing species that contains at least three atoms/molecule. In view of this and of (i), CIOz seems the only likely absorbing species. (iii) The band systems have a very low appearance pressure and any impurity, present in small amounts, would have to be an extraordinarily strong absorber to produce the observed systems. (iv) The three band systems are all observed on plates that also show the known bands of C102 in the near ultra-violet.C102 molecules must therefore have been present in the spectrograph in considerable concentration at the time of photographing the new band systems. Photolysis can hardly have been very extensive. In any case, photolysis could only affect molecules in the light beam (whose volume is small compared with the total volume of the spectrograph) and diffusion out of the beam would tend to keep the concentration of products small. (v) The band systems increase in intensity as more C102 is added to the spectrograph. Assignment of the systems to a photolysis product would therefore only be plausible if the product responsible were sufficiently inert to survive the time (a few minutes) from one exposure to the next.(vi) The band systems were observed with two different samples of C102. (vii) If the far ultra-violet band systems are not due to C102, it is surprising that C102 should show no absorption as strong as the strongest bands of the near ultra-violet system until quite short wavelengths (5 1400 A) are reached. According to Dibeler, Reese and Mann 6 the first ionization potential of C102 is 11-1 eV. We should certainly expect strong Rydberg transitions between 2000 and 1400 A. Moreover, we expect these Rydberg transitions (which partially remove an electron from an antibonding orbital) to be discrete, since even the near ultra- violet transition (where an electron is transferred from a non-bonding orbital to an anti-bonding one) is discrete.In addition, the magnitudes of the vibrational frequencies involved in the three electronic systems are all plausible for C102 as the absorbing species; and may be linked, as we show below, with the observed high intensities of the transitions. The ground state fundamental frequencies of C102 are known 7 to be 943 cm-1 (vial, symmetrical stretching), 1 1 10.5 cm-1 (v&, asymmetrical stretching) and 445 cm-1 (vzal, bending). 1020, 1011 cm-1 in the first system and 1051, 1091, 1068 cm-1 in the second system only plausibly represent, therefore, the symmetrical stretching frequencies of the upper states.They are increased from the ground state value. Now any intra-valency shell transition of C102 is expected to increase the Cl-0 length and reduce v1 (see Walsh 8). Decrease of v1 on electronic excitation is indeed observed in the near ultra-violet band system which is undoubtedly an intra-valency shell transition.8 At least the first two of the three new electronic transitions must therefore be Rydberg transitions. The wavelengths at which the transitions occurC. M. HUMPHRIES, A . D. WALSH AND P . A . WARSOP 141 accord with this conclusion. In a Rydberg transition associated with the first ion- ization potential, an electron is removed from the anti-bonding bl -EU molecular orbital and placed in a large Rydberg orbital wherein it is expected to be without considerable effect on the dimensions of the molecule.v1 should therefore be in- creased by the high energy excitation, as observed in the first and second band systems. The conclusion that these two systems are Rydberg may be linked with their high intensities, since the early Rydberg transitions are expected to be of high intensity. Emax. for the near ultra-violet system is reported to be at least 2000.9 It follows from the appearance pressure data that the far ultra-violet systems also have Emax. of at least the same order of magnitude. Rydberg transitions have Emax. values up to at least 104. That v; is not very different in the first and second systems is expected because the upper states of Rydberg transitions leading to the first ionization potential should each approximate to the ground state of the ion ClO;.vi is, in fact, some- what greater for the second than for the first system and so the Cl-0 length is presumably somewhat less in the upper state of the second system than in the upper state of the first. That the bands of the main progression in the second system fall off in intensity more slowly than do the bands of the first system accords with v; -v’i being greater for the second system than for the first. Indeed, the association of the slower falling off in intensity with the higher value of v; constitutes in itself an argument that both band systems are due to a molecule whose v‘; frequency is less than the v j frequencies of ca. 1020 and ca. 1050 cm-1; and so there is a further argument consistent with C102 being the absorbing species.The frequency of ca. 520cm-1 in the second system must represent vi. The weakness of bands E, G and J relative to bands D, IF and H in the second system, and the non-appearance of bands showing a separation of ca. 500cm-1 in the first system, then implies that in both these systems the electronic excitation causes a much more important change in the Cl-0 distance than in the OClO angle. This is in accord with the diagram plotted by one of us 8 correlating the molecular orbitals of linear and bent C102; an electron in the bl-EB orbital has only a minor effect on the OClO angle. Accepting the first two band systems as the first allowed Rydberg transitions, the upper orbital of the 1829 A transition may be labelled (xul), while the upper orbital of the 1628 A transition may be labelled (pal) or (pbl).All the bands should therefore be perpendicular in type, in contrast to the parallel bands of the near ultra-violet system. The far ultra-violet bands are observed to be narrower than the near ultra-violet bands. The third new band system has about the same intensity as the first two transi- tions and is therefore probably a Rydberg transition also. The frequency of ca. 500 cm-1 must represent vi. On the other hand, the third system differs from the first two systems in that (a) the first band is not the strongest, (b) a considerably larger number of bands apparently forming a progression in vi is present, (c) the intensity of the (OlO)+(OOO) band is far higher relative to that of the (OOO)+(OOO) band.If the arrangement shown in column (a) of table 2 is adopted, v[ is apparently not excited and a considerable progression in v;, reaching maximum intensity in the third band, is present. If the arrangement shown in column (b) of table 2 is adopted, the frequency of ca. 1000 cm-1 must represent vi. In that case, v i is less in the third system than in either of the first two systems and any progression in v; would be expected to fall off in intensity even more rapidly than in the first system. Since, in fact, band L is more intense than band I, L cannot be simply the (lOO)+-000) transition, but must involve a considerable contribution from the (020) +(OW)142 SPECTRUM OF CHLORINE DIOXIDE transition. Similarly, comparison of fig. l(a) and l(b) suggests that band N must largely represent the (04O)t(OOO) vibronic transition of the third system.In cther words, we confirm the conclusion already reached that, however arranged, the bands of the third system involve a progression in v;. In contrast, v i does not appear in the first system and is only represented by very weak bands in the second system. From the vi considerations, the change in Cl-0 length brought about by the elec- tronic excitation is less in the third system than in the first and second systems; while, from the vi considerations, the change in OClO angle brought about by the excitation is much more profound in the third than in the first two systems. It is difficult, therefore, to suppose that the third system represents a Rydberg transition associated, like the first two systems, with the first ionization potential.The transition, however, does have the characteristics that might be expected for a Rydberg transition involving excitation of an electron from the a1 -FU orbital, i.e., a Rydberg transition associated with the second ionization potential. The nu orbital is C1--0 antibonding; but, by bending, the antibonding nature of the a1 component of %a is partially relieved and, especially at angles not too far removed from the ground state value of 116.5*,* the a1 -Zu orbital is expected to be largely a lone pair or non-bonding orbital.8 Removal of an electron from the al-n, orbital is ex- pected to cause a considerable increase in OClO angle (and hence a long progressiofi in the upper state bending frequency), but only a small change in C1-0 length (and hence little or no excitation of v;).Since the first excited state of the ion ClO$ differs from the ground state of NO2 by having an electron in the bl--Ku orbital, it seems probable that the OClO angle in the upper state of the third new band system is a few degrees less than the value (134") of the ON0 angle in the ground state of N O 2 . Assignment of the third new band system as the first Rydberg transition leading to the second ionization potential implies that (i) the upper orbital of the 1568 A transition is (sal) in type and the bands of the system are perpendicular; (ii) the upper orbitals of the 1568 and 1829 A transitions are indeed the same and, therefore, that (iii) the C102 molecule should possess an intra-valency shell transition, formulated as whose maximum intensity should lie at roughly the separation of the positions of maximum intensity of the 1829 and 1568 A transitions, viz., at ca.64,478 - 54,689 = 10,089 cm-1, which corresponds to ca. 10,000 A. A search for an absorption system in this difficult region of the infra-red may therefore provide a test of the assignments made here. The bands of transition (1) should be perpendicular in type and the transition should cause increases in both apex angle and C1-0 length. The near ultra-violet bands are satisfactorily assigned to the transition - * (a1)(b1)2, 2AlC. - @1)2(bl), 2B1, (1) * (a2)(b2)2(a1)2(b1)2, 2f42+. ' - (a2)2(b2>2(a1)2(b1>, 2B1. (2) Transitions (1) and (2) are the only allowed, low-lying transitions expected to possess discrete structure. Transitions involving transfer of an electron to the 21 -Zg orbital may well be responsible for the continuous absorption observed.It remains to discuss the alternation in spacing observed for the bands of the third far ultra-violet system. A possible explanation of this is that it is due to Fermi resonance. One would expect, from our assignment of the third system, that vi would be somewhat greater than v'i though less than v{ in the first and second systems. * from electron diffraction data.10 A combination of electron diffraction and infra-red data leads to the value 118.5".C. M. NUMPHRIES, A. D. WALSH AND P. A. WARSOP 143 v { may well therefore be - 1000 cm-1 as shown in arrangement (b) of table 2 ; and since v$ is -500 cm-1, Fermi resonance between vi and 2v5 is possible.The bands seem best regarded as primarily a v i progression ; but with some excitation of v i , because of Fermi resonance, in all bands from the third onwards. Data on the ground state levels of C02 (where v1 is well known to be in Fermi resonance with 2v2) show that it might be possible, because of Fermi resonance, for a vibrational progression to show an apparently alternating spacing. Thus, taking the highest energy level of each Fermi resonance group of levels for C02 11 an alternating spacing is evident : cm-1 0 667 1388 2077 2797 3502 4224 )667 )721 )689 )720 )705 )722 In a similar way, if particular members of each Fermi polyad are the most intense and the others are not resolved or are too weak to be observed, it might be possible to explain the alternating spacing found for C102.Reversing the argument, if Fermi resonance is accepted as the explanation of the alternating spacing, then v1 is to some extent increased by the electronic excitation of the third band system and we have a further argument that the third system is Rydberg in type. Fermi resonance may also play a part in determining the spacings of the bands in the 1628 system. The spacing between bands F and M is definitely greater than that between bands D and F, in spite of the expectation that in a v1 progression the spacing should decrease, rather than increase, as successive quanta of v1 are added. Further, ca. 520 is not far from ca. 1050/2 and there appears to be a short, long, short alternation in the spacings of bands D F H K’. It may be that the bands of the 1628 A system are best regarded as primarily a v; progression ; but with some excitation, due to Fermi resonance, of v$. 1 Coon, J. Chem. Physics, 1946, 14, 665. 2 Coon, Physic. Rev., 1952, 85, 746. 3 Coon and Qrtiz, J. Mol. Spectr., 1957, 1, 81. 4 Price and Simpson, Trans. Faraday Soc., 1941, 37, 106. 5 Bray, 2. physik. Chem., 1906, 54, 575. 6 Dibeler, Reese and Mann, J. Chem. Physics, 1957, 27, 176. 7 Nielsen and Woltz, J. Chem. Physics, 1952, 20, 1878. 8 Walsh, J. Chem. Soc., 1953, 2266. 9 Goodeve and Stein, Trans. Faraday Soc., 1929, 25, 738. 10 Dunitz and Hedberg, J. Amer. Chem. Soc., 1950, 72, 3108. 11 Taylor, Benedict and Strong, J. Chem. Physics, 1952, 20, 1884.