Absorption Spectra of the Hydrides, Deuterides and Halides of Group 5 Elements BY C. M. HUMPHRIES, A. D. WALSH AND P. A. WARSOP Chemistry Dept., Queen’s College, Dundee Received 14th January, 1963 Data are reported on the absorption spectra of PH3, PD3, AsH3, AsD3, SbH3, PF3 and PCl3 and compared with corresponding data for NH3, ND3. All the electronic transitions observed are Rydberg transitions leading to the first ionization potentials; and all with discrete structure are represented by long progressions in the upper state bending frequency (v2) alone. The regularity of all the progressions observed is compatible with the upper states of the molecules being (i) planar; or (ii) pyramidal with such a low barrier to inversion that all the transitions observed are to vibra- tional levels well above the top of the barrier; or (iii) pyramidal with such a high barrier to in- version that all the transitions observed are to vibrational levels well below the top of the barrier.However, consideration of the energies required to change the inter-bond angles in the PF3 and PCl3 molecules from their excited-state values to their ground-state values leads to the conclusion that whereas (i) applies to the Rydberg states of NH3 (and the ground state of the NHS ion), (iii) applies to the Rydberg states of PF3 and PCl3 (and the ground states of the PF; and PClf ions). It is stressed that whereas the spacing of the observed bands for NH3, ND3, PF3 and PCl3 is -v;. the spacing of the observed bands for PH3, PD3, AsH3, AsD3 and SbH3 is -v;/2. The contrast between v;/vl for the latter group of molecules and for PF3, PC13 on the one hand and NH3, ND3 on the other is used as an argument that neither (iii) not (i) applies to the upper states of PH3.. . SbH3 and that these molecules probably have Rydberg states (and positive ion ground states) described by (ii). A previous paper 1 dealt with the ultra-violet absorption spectrum of ammonia. Five separate electronic transitions were recognized, each of Rydberg type and each giving rise to a long progression of bands in the excited state vibration (v;) that causes a symmetrical change of the HNH angles. Starting with the longest wavelength transition, we shall refer to these five transitions as the first, second, . . . transitions of ammonia. The origins of the first, third and fourth transitions lie at 2168, 1434 and 1330A.The origin of the second transition was thought to lie at 1665 A, but has since been shown by Douglas and Hollas 2 to lie one quantum of vi to longer wavelengths, i-e., at 1689A. The first four transitions were each proved,l by a vibrational analysis, to have planar upper states. Douglas and Hollas 2 have since confirmed, by a rotational analysis, that the second transition has a planar upper state. The upper state of the Mth transition (of unknown origin, but represented by bands in the neighbourhood of 1268A) and, indeed, the upper states of all the Rydberg transitions of ammonia (including the ground state of the NH; ion) were concluded to be also planar. Certain weak parallel bands observed by Douglas and Hollas 2 between 1546 and 1458 A may represent a further electronic transition or possibly a v; + nv; progression belonging to the second (1 689 A) transi- tion. Other Rydberg transitions must occur as the ionization potential at 1221 A is approached.The stronger bands of the second transition are perpendicular and the upper state electronic symmetry is E”.2 The upper state of the first transition has A; electronic symmetry 1 and the bands observed should be parallel. In the present paper the absorption spectra of PH3, PD3, AsH3, AsD3, SbH3, PF3 and PC13 are discussed. All the spectra were photographed with the aid of a 2-m vacuum spectrograph. The dispersion was ca. 8-7 A/=. 148C. M. HUMPHRIES, A . D. WALSH AND P. A. WARSOP 149 RESULTS The full data for the hydrides and deuterides will be published elsewhere; a preliminary account has already been given.3 The spectra of deutero phosphine and phosphorus trichloride are reproduced on plate I ; the spectrum of phosphorus trifluoride is shown on plate 11.The lowest energy absorption transition (see plate 11) of phosphorus trifluoride consists at low pressures, of two very diffuse peaks, one at 1564 A and one at 1515 A. TABLE TH THE 1405 A PROGRESSION OF PHOSPHORUS TRIFLUORIDE frequency (cm-I) Av (cm-1) ca. 71,174 ca. 71,661 72,144 72,602 \ca. 487 ca. 483 73,958 75,360 76,717 77,174 > 456 77y6307 463 78,093 -< 467 78,560 \ 452 79,012- > 457 79,455 - TABLE THE 1212 A PROGRESSION OF PHOSPHORUS TRIFLUORIDE frequency (cm-1) Av (cm-1) 82,511 7 470 472 82,981 ,-< 451 459 456 474 83,453 / 83,904 -< 84,363 -< 84,819 ___( 85,749 85,2g33 456 \ 465 \ ca.460 ca. 457 4 463 451 462 86,214 2 ca. 86,674 87,131 87,594 -< 88,045 / 88,507 A As the pressure is increased, these rapidly merge into each other and the resulting continuum spreads out slowly and symmetrically to long and short wavelengths. At high pressures, the short wavelength side of the continuum overlaps a long pro- gression of rather diffuse bands which first increase and then decrease in intensity150 SPECTRA OF HYDRIDES, ETC., OF GROUP 5 ELEMENTS as they proceed to short wavelengths. The first observed band is at 1405A and maximum intensity occurs around 1300A, i.e., at about the thirteenth band. As may be seen from plate 11, the envelope of this second, banded absorption region resembles, but has a lower peak intensity than, the envelope of the first, continuous absorption region.The short wavelength side of the second absorption region TABLE 3.-THE 1591 A PROGRESSION OF PHOSPHORUS TRICHLORIDE frequency (cm-1) 62,823 63,089 63,361 63,619 63,879 64,152 4) 64,680 64,411 2’ \ :: > \ 64,931 -------< 65,197 65,464 65,734 < 65,985 ~ 66,231 66,478 - > i 66,778 67,025 Ar(cm-1) 266 272 258 260 273 269 269 25 1 266 267 270 25 1 246 247 300 247 overlaps a further progression of diffuse bands beginning at 1212A. These further diffuse bands increase in intensity as they proceed to short wavelengths; they can be followed up to about the fourteenth band (at ca. 1130A), but thenceforward merge into strong continuous absorption.The third absorption region is somewhat stronger than the second and similar in intensity to the first absorption region. Frequency measurements of the centres of the bands in the 1405 and 1212 A progressions are given in tables 1 and 2 respectively. The diffuse- ness of the bands and the occasional presence of emission lines in the Lyman continuum used as background make precise measure- ment impossible and imply that no stress should be put upon the deviations from constancy of the band separations in the two progressions. The mean band separation in the 1405 A progression is 460 cm-1 and in the 1212 A progression is 461 cm-1. The transitions are so strong that they must be allowed. Assuming the transitions do not reduce the three-fold symmetry of the ground state, then the frequencies involved must be totally symmetrical with respect to C3v sym- metry, i.e., v; (symmetrical P-F stretching) or vi (symmetrical bending).v; and v; are, respectively, 892 and 487 cm-1.4 The mag- nitudes of the frequencies involved then leave no doubt that the frequencies represent v;. One notes that v2 is but little reduced by the electronic excitation. The lowest energy absorption region of phosphorus trichloride consists of a continuum of maximum intensity around 1750 A. With increasing pressure, this continuum spreads to both red and violet and on the violet side overlaps the second electronic transition. The latter consists of a long progression of diffuse bands which mount in intensity and, after the seventeenth band, merge into continuous absorption whose intensity is a maximum around 1470A and fa, 11s off towards shorter wavelengths.The first observed band is at 1591 A. The measured fre- quencies and frequency differences of the bands are given in table 3. The mean band separation is 263 cm-1. The frequency involved is undoubtedly vi; IJ; is 240 cm-1.5 Again one notes that v2 is changed very little by the electronic excitation. The difficulties of measurement again mean that the deviations from constancy ofPLATE 1 .-The absorption spectra of (a) trideutero phosphine, (b) phosphorus trichloride. PLATE 2.-The absorption spectrum of phosphorus trifluoride. [To face page 150.C. M. HUMPHRIES, A. D. WALSH AND P . A. WARSOP 151 the band separations shown in table 3 are not necessarily significant.To short wavelengths of the 1591 A progression and associated continuum, is a further con- tinuum, the intensity of which rises rapidly to a maximum and then decreases gradually. This continuum has the appearance of a very broad, violet-degraded, band and has maximum intensity at 1385 A. At the pressures used (up to ca. 1 mm Hg in a 4 m path), the lowest energy ab- sorption transition of phosphine or deutero phosphine (see plate I) consists of a continuum of maximum intensity at 1800 A.* On its short wavelength side, this continuum overlaps a long progression of bands which has a higher appearance pressure than does the peak of the continuum. The PD3 bands are sharper than those of PH3. Nine PH3 bands have been measured and fit the formula Twelve PD3 bands have been measured and fit the formula Because of the overlap by the 1800 A continuum, the first observed band with either PH3 or PD3 is not necessarily the origin of the progression.On the short wave- length side, each progression is overlapped by further absorption. The observed bands of progressions (1) and (2) do not change markedly in intensity with wave- length and the overlap with other (stronger) electronic transitions to both long and short wavelengths makes it impossible to determine the position of maximum intensity. The further absorption that overlaps the short wavelength side of the second electronic transition consists of several overlapping progressions. With PH3, the bands are very diffuse and it is impossible to disentangle the various pro- gressions with certainty.The first observed band lies close to 15OOA and band separations increase from about 420cm-1 at the long wavelength end to between 500 and 600 cm-1 at shorter wavelengths. The first PD3 band which does not belong to progression (2) was earlier 3 observed at 69,716 em-1, but bands have now been traced back to 67,532 cm-1. The band separations are, initially at least, ca. 400 cm-1. A further progression of 14 PD3 bands fits the formula Each of the progressions observed for PH3 and PD3, like all those observed for NH3, is undoubtedly a progression in vh. v; has the values 992 and 730 cm-1 respectively, for PH3 and PD3.9 It thus appears that all the electronic transitions observed for PH3 and PD3 roughly halve the value of v2.The spectra of AsH3 and AsD3 are very similar to those of PH3 and PD3, but are more diffuse. The first, continuous absorption region has a maximum at about 1830 A. The second electronic transition, with AsH3, is represented by a progression of twelve bands obeying the formula and, with AsD3, by another progression of twelve bands obeying the formula v (cm-1) = 62,801 +488.0 n’+7.84 n’2 (n’ = 0, 1, 2, . . ., 8). v (cm-1) = 62,865 + 361.6 n‘ + 4.0 n’2 (n’ = 0, 1, 2, . . ., 11). (1) (2) v (cm-1) = 74,946+356-4n’+2.11 12’2 (n’ = 0, 1, 2,. . ,, 13). (3) v (em-1) = 62,453-1-421.5 n‘+5.94 n’2 (n’ = 0, 1,2, . . ., ll), v (cm-1) = 63,057 + 319.0 n’ +2.70 n’2 (n’ = 0, 1,2, . . ., 11). (4) (5) * Using 760 mm Hg of phosphine (prepared from pliosphonium iodide and alkali) in a path length of 110 cm, Cheesman and Emelbus 6 observed two very faint, diffuse bands, at 231 5 and 2290 A.The bands were independently observed by Melville.7 Since the separation of the bands is 470 cm-1 (i.e., of the same magnitude as the band separations in the shorter wavelength transitions), it seemed plausible that the bands were due to phosphine and not to an impurity. Dr. L. Mayor, however, working in this Department and using phosphine prepared from magnesium aluminium phosphide, has been unable to reproduce the bands.152 SPECTRA OF HYDRIDES, ETC., OF GROUP 5 ELEMENTS Both progressions are overlapped at their long wavelength end by the stronger 1830 8, continuum ; and so the first observed bands may not be the progression origins.A further progression of AsD3 can be recognized; it obeys the formula v (cm-1) = 77,098+313-6 n'+2.43 n'2 (n' = 0, 1,2, . . ., 9). (6) Between progressions (5) and (6) of AsD3, fragments of at least two other progres- sions, with band separations - 350-400 cm-1, can be recognized ; and there are signs of AsH3 progressions to short wavelengths of progression (4). The band separations in all these progressions of AsH3 and AsD3 undoubtedly represent vi. vi has the values 906 and 660cm-1 respectively for AsH3 and AsD3.10 All the banded elec- tronic transitions observed thus result in (roughly) halving v2. The first transition of stibine is represented by a continuum of maximum intensity at about 1970A; and the second by a progression obeying the formula v (cm-1) = 58,320+419-4 n'+4-94 n'2 (n' = 0, 1,2, .. ., 7). (7) Very diffuse bands to short wavelengths of this progression represent a third elec- tronic transition. Still further to short wavelengths occurs a transition represented by a progression obeying the formula, v (cm-1) = 77,395+419 n'-4*4 n'2 (n' = 0, 1, 2. . . ., 7). vi for SbH3 is 796 cm-1.11 It follows that, with SbH3 as with PH3, PD3, AsH3 and AsD3, the banded electronic transitions observed result in a rough halving of v2. DISCUSSION The halide, hydride and deuteride spectra described all have considerable similarities. In each spectrum, the second electronic transition is weaker than the first and also weaker than some of the shorter wavelength transitions. The first electronic transition in each spectrum is predissociated; with NH3 weakly, so that vibrational structure is retained and only rotational structure is lost ; with the other molecules so strongly that all structure (except in the spectrum of PF3, the two diffuse humps at 1564 and 1515 A, whose interpretation is unknown) is lost.Each discrete transition is represented by a long progression in v; and there is no evidence of any excitation of v;. Excitation of long vi progressions implies that the electron concerned comes from the lone pair orbital (a1 with respect to C3v symmetry) on the group 5 atom, which orbital is the most weakly bound orbital occupied in the ground state and is approximately non-bonding. It is the presence of two electrons in the a1 lone pair orbital that causes the ground states to be pyramidal.The absence of excitation of v; implies that each transition causes little change in the bonding between the central atom and the other atoms ; and, therefore, that the electron concerned passes to a Rydberg orbital, which is the only type of orbital, unoccupied in the ground state, that is expected (because of its size) to be without considerable influence on the molecular dimensions. All this is made clear by the diagram plotted by one of us 12 correlating the orbitals of a pyramidal AH3 (or AB3) molecule with those of the same molecule in a planar configuration. The upper states may thus all be described as Rydberg states leading to the first ionization potential; and the properties of the various upper states for any one molecule should differ little from state to state and from the ground state of the ion AH: or ABZ.Clearly, too, the inter-bond angles in all the upper states and in the ions must be considerably greater than in the ground states. With noC. M. HUMPHRIES, A . D . WALSH AND P. A . WARSOP 153 transition of any of the molecules we are dealing with have we observed alternate long and short spacings between the bands. This is consistent with one of three possibilities. The first is that all the observed bands for any particular molecule represent transitions to a planar upper state. The second is that all the observed bands represent transitions to vibrational energy levels lying well below the top of the barrier to inversion of a pyramidal state. The third is that all the observed bands represent transitions to vibrational energy levels lying well above the top of the barrier to inversion of a pyramidal state.With NH3 we have already con- cluded that the upper states and the ion NH< are planar. It does not, however, necessarily follow from the similarities between the spectrum of ammonia and the spectra of all the other molecules studied that with every molecule the inter- bond angle in the upper states is increased by the electronic excitation as far as 120", i.e., that the first possibility holds. In any particular excited state of a molecule AB3, the energy required to bend the BAB angles from their equilibrium value in the excited state to their value in the ground state is given by the separation of the origin and the position of maximum intensity in the relevant transition.Thus, with the first Rydberg transition of NH3, maximum intensity occurs 1 at about v' = 6 (51,550 cm-1) and the origin certainly at 2168 A (46,181 cm-1) ; so that the energy required to bend the excited state from planarity to a pyramidal form in which the HNH angle has the ground state value of ca. 107" is about 5400cm-1. With the second Rydberg transition of NH3, maximum intensity 1 occurs at about v' = 5 (64,981 cm-1) and the origin 2 certainly at 1689 A ; so that the separation is about 5800 cm-1. The rough equality of the two calculated energies agrees with the expectation that all the Rydberg states should have properties that are very similar and approximate to those of the ion. One would expect that the bending energy would be rather more in the second than in the first excited state, since the vibrational frequency (and hence the bending force constant) are somewhat greater in the second than in the first excited state; the above rough estimates agree with this.For the 1434 and 1330 A transitions, ac- cepting the origins and positions of maximum intensity given previously,l the calculated energies are rather less ( N 3800 cm-1). The vibrational frequencies, and hence the bending force constants, would lead one to expect bending energies greater than that for the first excited state; and there may be an indication here that the true origins of the third and fourth transitions lie one or two quanta of v;l to the red of the positions hitherto given, or more probably that the positions of maximum intensity (which are difficult to determine with precision) lie one or two quanta of v; to the violet of the positions previously given.It is noteworthy that if one adds two quanta of v; to the previously given separations of the origins and positions of maximum intensity, one obtains bending energies that are - 5600 and - 5700 cm-1 for the upper states of the 1434 and 1330 A transitions respectively. The bending energies are then all very similar and their small variation qualitatively accords with the variation in vi frequency and deduced variation in bending force constant. Thus the difference between the adiabatic and vertical first ionization potentials may be -0.7 eV rather than the previously given value of -0-5 eV. With PF3, the separation of the first observed band and the position of maximum intensity of the second electronic transition is about 5500 cm-1; while the separations of the first and last observed bands in the third electronic transition is 5996 cm-1. These data and the apparently similar envelopes of the first and second transitions are consistent with the expectation that the FPF angle in each upper Rydberg state is, at least roughly, the same; and with the first observed bands of the second and third transitions lying within a few quanta of 460 cm-1 from the origins.With154 SPECTRA OF HYDRIDES, ETC., OF GROUP 5 ELEMENTS these assumptions, the energy required to bend, without change in P-F length since v; is not excited, the FPF angle from its equilibrium value in the Rydberg states of PF3 (and in the ion PF;) to its value in the ground state is within a few multiples of 460cm-1 from 5500cm-1.In other words, the difference between the vertical and adiabatic first ionization energies of PF3 should be ~ 0 - 7 eV. Let us assume that the upper states of the observed transitions of PF3 are planar. The valence force expression for the potential energy V of a PF3 molecule that is planar in its equilibrium form, as a function of (i) changes (Q) in the P-F distances, (ii) changes (6)in the FPF angles (iii) changes (A) in the angle between a PF line and the F3 plane, is (ref. (13), p. 178) : 2V = ki<Q?2 + Q?3 + Q?4> + k d d k + 6224 + 6 L ) + k~(A?2 +A?3 +A:4). kl, Icb and kA represent the appropriate force constants. Al~0,13 4n2v;c2 = I+-- -2, ( ::)n: where l is the P-F length and the other symbols have obvious significance. v; is not excited in the observed transitions, we may take l as approximately the same as the ground state P-F length.Electron diffraction and microwave data 1.1 are consistent with the ground state FPF angle being 100" and P-F length being 1-53(5) A. The energy required to bend symmetrically, without chaFge in P-F length, the supposedly planar state to a pyramidal form in which FPF = loo", should, from eqn. (9) and (10) be given by where 6 = 20/57-3 radians and, since FfiF = 100" corresponds to an angle of 28" between an F-P line and the F3 plane, A = 28/57-3 radians. To evaluate kb for the planar molecule requires a knowledge of v; and vi, which are unknown.How- ever, the first term on the right-hand side of eqn. (11) is expected to be positive and the second term alone leads, on inserting numerical quantities including vi = 460 cm-1, to an energy of ca. 36,000 cm-1. Since the calculated energy is so much greater than ca. 5500cm-1, then it is improbable that the Rydberg transitions of PF3 lead to planar upper states. Since the separation of the first and last bands observed in the second electronic transition of PF3 is only 79,455-71,174 = 8281 cm-1, then all the vibronic transi- tions observed take place to upper state vibrational levels that probably lie well below the top of the potential barrier to inversion; and the absence of any measurable inversion effects is understandable. Assuming then pyramidal upper states, we may estimate the FPF angle in these states as follows.The energy required to bend the three FPF angles in a pyramidal PF3 molecule by an amount 6 without change in the P-F distances should be given approximately by the expression v = $kad2, (12) (ref. (13), p. 175). k6/P in the ground state is known, from the v; and v; frequencies, to be 1-15 x 105 dynes/cm (ref. (13), p. 177). As above, we shall take l in the excited state to be the same as in the ground state. The observed values of vi are prac- tically the same (460 cm-1) and only slightly less than the value of v; (487 cm-1). The force constants controlling bending may, therefore, be taken as approximatelyC. M. HUMPHRIES, A . D . WALSH AND P. A . WARSOP 155 the same in the ground and excited states and more precisely as in the ratio kL/k: = (460/487)2.Inserting the value V = 5500 cm-1 in eqn. (12),* we then find that the equilibrium FPF angle in the Rydberg states should be, roughly, only 110". Very similar arguments and conclusions apply to PC13. Here microwave data yield an angle of 100" and a bond length ( I ) of 2.04(3) A in the ground state ; 15 while force constant calculations, based on v; and v;, yield ka/P = 0.43 x 105 dynes/cm for the pyramidal ground state. as the position of maximum intensity and assuming the 1591 A band to lie within a few vi quanta of the origin, one obtains ~ 5 2 0 0 cm-1 for the difference between the vertical and adiabatic first ionization energies and an upper-state equilibrium angle of roughly 11 lo.* Similar calculations and arguments cannot be applied to PH3, AsH3 and SbH3 because for no transition of these molecules is the separation of the origin and the position of maximum intensity known with sufficient certainty. However, a major reason for Rydberg excitation of PF3 and PC13, unlike NH3, failing to produce planar molecules is that the ground state inter-bond angles are considerably smaller than in NH3.It would, therefore, not be surprising if Rydberg excitation of PH3, AsH3 and SbH3 (which molecules have ground state inter-bond angles of only 93" or less) failed to produce planar upper states. In this connection, however, although the lowest energy, excited state of the NH2 free radical is linear, that of the PH2 radical is probably non-linear.16 On the other hand, it would seem likely that Rydberg excitation of PH3, AsH3 and SbH3 will produce molecules with inter-bond angles at least approaching the value of the HNH angle in the ground state of NH3; and therefore that the upper states should show inversion effects.These have not been observed, but may be present in thus far unobserved transitions. If so, it is only likely that they occur in transitions close to the (unobserved) origins ; if they occurred in transitions of higher energy than the shortest wavelength members of the various progressions, the energy range covered by the observed progression bands would imply barriers to inversion much higher than the known barrier height in the ground state of NH3. In this connection also, there is one outstanding difference between the transitions of NH3, ND3, PF3 and PCl3 on the one hand and the transitions of PH3, PD3, AsH3, AsD3 and SbH3 on the other (see ref. (3) for a discussion, confined to the hydrides, of the difference).With each of the former molecules, the values of v; in the various transitions are all of about the same magnitude as the value of vi. [For NH3, vl here refers to the separation of the (slightly split) levels well below the top of the barrier to inversion.] With each of the molecules PH3, PD3, AsH3, AsD3 and SbH3, however, the observed values of vi are close to one-half the value of vi. The contrast between the latter molecules and PF3, PC13 is explicable if we assume that the upper states of PH3, etc., differ from the upper states of PF3, PCl3 in having a barrier to inversion that is of low or zero height ; and if we assume that the observed transitions of PH3, etc., are all to levels well above the top of any inversion barrier, whereas the observed transitions of PF3, PCl3 (as we have concluded) are all to levels well below the top of the barrier.For any particular electronic state, the spacing of the vibrational levels well above the top of the barrier to inversion should be approximately one-half of the spacing of the levels well below the top of the barrier. If, therefore, electronic excitation changes but little the vibration frequency vz except For the second electronic transition, taking 1470 * The interval between the estimated position of maximum intensity and assumed origin only enters these calculations as the square root; even if the interval used is in error by 50 %, the cal- culated angle is only changed by 2-t".156 SPECTRA OF HYDRIDES, ETC., OF GROUP 5 ELEMENTS in so far as inversion effects are concerned (as seems true for PF3 and PC13), the fact that the observed bands of PH3, etc., have a spacingv;/2, whereas the observed bands of PF3, Pc13 have a spacing -v; is understandable. The difference between the two groups of molecules is that for PF3 and PC13 we have observed only transi- tions to levels below the top of the barrier to inversion whereas for PH3, etc., we have observed only transitions to levels above the top of any barrier.Conversely, one may argue that the halving of frequency for PH3, etc., constitutes evidence that the excited states of PH3, etc., do not possess a high barrier to inversion and, unlike PF3 and Pc13, are not simply pyramidal.If the excited states of PH3, etc., were simply pyramidal, one would have to make the unplausible assumption that corresponding electronic excitation makes little difference to the force constant controlling bending in PF3, Pcl3 but reduces the force constant for PH3, etc., to one-quarter of its ground state value. We conclude, then, from the contrast of PH3, etc., with PF3, PC13, that the ex- cited states of PH3, etc., are either strictly planar or pyramidal with a low barrier to inversion. In themselves, the observed spectra of PH3, etc., are consistent with either possibility. However, the contrast between PH3, etc. (where v; - vi/2) and NH3 (where v;-vi and the upper states are certainly planar) then constitutes an argument that the upper states of PH3, etc., are not completely planar and, by elimination, must be pyramidal with a low barrier to inversion.If they were com- pletely planar, the striking difference between the upper state spacings of PH3, etc., and those of NH3 would be correlated with no other outstanding difference and it would seem particularly difficult to understand the contrast in values of v&. To this argument we may add the points made above concerning the very large increases of angle needed to make PH3, etc., planar and concerning the contrast of PH2 and NHz. Admittedly, it is not easy to see why, for NH3, vi should be ~ v ; . Since the electronic excitation apart from inversion effects does not greatly change the v2 vibration frequency in PF3, in Pc13 and (as is necessary for our explanation of why v;-v;/2) in PH3, etc., one might expect that with NH3 v; would be 4 2 .It seems that once there is no double minimum in the potential function and the excited states are strictly planar, some new effect comes in that roughly doubles the vi frequency in the excited states of NH3 from its expected value. As we have argued before,3 the dificulty lies with NH3. One possible explanation is as follows. The lone-pair electron which is excited in all the electronic transitions dealt with here has a tendency to bend the molecule from a planar to a pyramidal shape.12 The removal of a lone-pair electron from a pyramidal molecule such as PF3 causes the bond angle to increase without apparently changing the vibrational frequency v2 very much.The removal of a lone-pair electron from a planar molecule (e.g., from NH> to form NH:+), however, cannot change the bond angles (see ref. (12)), but might be expected to make it more difficult to bend the molecule, i.e., the vibrational frequency might be expected to increase when no change of shape is possible. NH3 in its ground state is so near to being planar that the removal of a single lone-pair electron may be sufficient not only to make the molecule planar (with a frequency 4-1~72) but also to cause the frequency to increase until it is almost equal to vi. In other words, all the observed spectra can be understood if (a) the removal of a lone-pair electron from a pyramidal molecule causes the bond angle to increase but, provided the molecule does not become planar, makes little change in the vibrational frequency v2; (b) the removal of a lone-pair electron from a planar molecule causes the frequency v2 to increase; (c) the Rydberg states of PF3, PCl3C. M. HUMPHRIES, A . D . WALSH AND P . A . WARSOP 157 are pyramidal with a high barrier to inversion; (d) the Rydberg states of PH3, PD3, AsH3, AsD3 and SbH3 are pyramidal with a low barrier to inversion ; (e) NH3, ND3 are planar in their Rydberg states. 1 Walsh and Warsop, Trans. Faraday SOC., 1961, 57, 345. 2 Douglas and Hollas, Can. J. Physics, 1961, 39, 479. 3 Walsh and Warsop, 4th Int. ConJ Molecular Spectroscopy, Bologna, 1959 (Pergamon Press, 4 Wilson and Polo, J . Chern. Physics, 1952, 20, 1716. 5 Davis and Oetjen, J. Molec. Spectr., 1958, 2, 253. 6 Cheesman and Emelkus, J. Chem. SOC., 1932, 2847. 7 Melville, Nature, 1932, 129, 546. 8 Thompson and Duncan, J. Chem. Physics, 1946, 14, 573. 9 McConaghie and Nielsen, J. Chern. Physics, 1953, 21, 1836. 10 McConaghie and Nielsen, Physic. Rev., 1949, 75, 633. 11 Haynie and Nielsen, J, Chenz. Physics, 1953, 21, 1839. 1 2 Walsh, J. Chem. SOC., 1953, 2296, 2301. 1 3 Herzberg, Infra-red and Ramaiz Spectra (Van Nostrand, New York, 1945). 14 Williams, Sheridan and Gordy, J. Chem. Physics, 1952, 20, 164. 15 Kisliuk and Townes, J. Chem. Physics, 1950, IS, 1109. 16 Ramsay, Determination of Organic Structures by Physical Methods, vol. 2 (Academic Press, Oxford, 1962). New York, 1962).