MICROWAVE SPECTRA OF DEUTERATED FURANS STRUCTWRE OF THE FURAN MOLECULE BY BBRGE BAK, LISE HANSEN AND JOHN RASTRUP-ANDERSEN Chemical Laboratory of the University of Copenhagen, Copenhagen Received 17th January, 1955 6- and P-monodeutero-, and a : a’-dideuterofuran have been prepared and their micro- wave spectra have been recorded and analyzed. Values of the rotational constants so obtained in connection with known values of the rotational constants of ordinary furan 1 are insufficient for a complete calculation of the 8 geometrical parameters of the molecule but additional reasonable assumptions concerning the length of the C-H bonds result in 5 molecular models among which a further choice may, be made by means of valence theory. This microwave work was undertaken in order to establish the molecular structure of furan with less ambiguity and uncertainty than has been achieved by electron-diffraction technique? Since furan is planar, the principal moments of inertia obey the relation la -1- I’ = I,, i.e.only two quantitative results are obtained which is quite insufficient to establish the magnitude of the 8 geo- metrical parameters of the molecule. Therefore, isotopic molecules must be studied. Various isotopic species are, however, of unequal value for the purpose, Generally, the most precise calculation of the molecular geometry follows from a study of species derived from the parent molecule by isotopic substitution of one atom. For that reason a- and P-monodeuterofurans were investigated, a : a’- dideuterofuran being included in the study only as an easy accessible means of control.In this way, 4 additional quantitative results were obtained. The total number of 8 necessary measured quantities now could have been procured by investigating isotopic species as, e.g., 12C4H4180, 12C313CH4160, etc., but the cost and the difficulties connected with the preparation of such compounds made this route impassable. 31111 order to state definite, possible models of furan, com- patible with the microwave measurements, assumptions as to the lengths of the C-H bonds were made. EXPERIMENTAL The a- and p-monomercurials (chlorides) and the a : a’-dimercurial (chloride) of furans were prepared according to Gilman and Wright.3 The yields reported by these authors were very satisfactorily reproduced.Mercury equivalent weights for the three compounds, found by boiling ca. 300 mg of the mercurial with 5 cm3 4MHC1, diluting with 20-30 cm3 hot water and precipitating Hg2+ as HgS before cooling, were, respectively, 300.1, 302.2, 283.0; calc. : 303.1, 303-1, 269.2. Potassium bromide discs were prepared from the three inercurials by pressing ca. 3 mg compound with 500 mg KBr4. The infra-red spectra of the discs, taken by a Beckmann I.R.2 instrument (700-4000 cni-1) showed that the mercurials were not contaminated with each other. Also, the absence of other im- purities seemed probable except in the case of a : a’-dimercurifuranylchloride where the low line intensity in the spectrum suggests that some spectroscopically rather inactive impurity is present (compare the Hg analysis).The conversion of these mercurials to deuteroderivatives took place by the following standard procedure. 0.065 mole DCL prepared from 12 g benzoyl chloride and 0.60 cm3 D20 was led into 24 cm3 ice-cooled D20, placed in a 100 cm3 flask, by means of a stream of dry hydrogen. In advance, 10 g mercurial (0.033, 0.033, 0.017 mole) had been placed in the same flask 30B. BAK, L. HANSEN A N D J . RASTRUP-ANDERSEN 31 which was now heated slowfy with a free flame to 100" C under continued flushing with dry hydrogen that passed a 30cm, air-cooled Vigreux column and a trap, immersed in liquid air, before leaving the system. Almost from the beginning the heating caused the evolution of (deutero) furan vapours which were separated from most of the water in the Vigreux column before being condensed in the trap.The connection between the column and the trap was now interrupted, and the latter was evacuated at about - 190" C . Under pressure control the contents of the trap were afterwards distilled at room temper- ature. The initial equilibrium pressure was approximately 400 mm. Fractions were taken until this pressure had dropped to 20mm, the water vapour pressure. The dis- tillate was dried for 30 min over anhydrous CaS04 and redistilled into a dry flask from which it was distilled in 25 cm3 fractions at constant temperature (ca. .20° C) and p = ca. 480 mm Hg. In this way we got 1.60 g a-deutero-, 1.10 g 13-deutero-, and 0-50 g a : a'-di- deuterofuran. The two first-mentioned distilled within the 470-490 mm pressure interval at 19.5-20*0" C while the dideutero sample, collected after three distillations when con- siderable " tails " were discarded, had p = 460-495 mm at 20" C.The yields are, there- fore, 70, 48 and 19 %. The infra-red absorption curves of the vapours, taken on a Beckmaiin I.R.2 instrument, showed that the three samples were neither contaminated with each other nor with ordinary furan (except for traces). The preparations have, therefore, been carried out without undesired exchange of H with D. This was fully confirmed by the following examination of the inicrowave spectra. These spectra were taken at room temperature in a conventional Stark modulation apparatus, working in the 17500-26000 Mcls interval, with a crystal-controlled frequency measuring system carefully calibrated against microwave lines of well-established frequency.Initially, Sirvetz's determinations of the line frequencies of ordinary furan 1 were checked and confirmed, Our observed and calculated microwave absorption frequencies are given in table 1 together with Sirvetz's results. This table does not include all observed frequencies ; a complete list would include about 1000 observations. TABLE 1 .-OBSERVED AND CALCULATED MICROWAVE ABSORPTION FREQUENCIES (in Mc/s) FOR FUR AN,^ a- AND 13-MONODEUTERO-, AND a : E'DIDEUTEROFURAN furan @-deuterofuran transition obs. (pa - lines) frequency 11.1321.2 23259.30+ 10.1320.2 23453*13+ 31.3+31.2 23352.476 42,3352.2 23305.88 53.3353.2 23213.45 64.3364.2 23055.80 ~ 75.3475.2 22810.92 86.33864 22458.99 97.3-fg7.2 2198430 108.3-+108.2 21377.91 1 1g.3-fllg.2 20637.71 1210.3-t 1210.2 1976798 30.3-332.2 23384.46 41.3343.2 23402.53 52,3354.2 23440.06 63.3t65.2 23507.71 74.3+76.2 23619.06 85.3437.2 23790.73 107.3+10g.2 24399.197 96.3+98.2 24043.08 ll8.3-+ll10.2 24888-97 42.2-+44.1 129.3-2.1211.2 25541 *64 53.2+55.1 calc.frequency 23259.29a 23453.1 3a 2335218 23305.76 23214.19 23056.96 228 1257 22461.20 21987.37 21 38 1.92 20642.69 19773-89 2338442a 23402.68 23440.46 23508.42 23620-09 23792.15 24044-82 24402.04 2489 1 *60 25544-73 * 8 centrifugal correction 0.01 0 0.29 0.12 - 0.74 - 1.16 - 1.65 - 2-21 - 3.07 - 4.01 - 4.98 - 5.91 0.04 - 0.15 - 0.40 - 0.71 - 1.03 -- 1.42 - 1.74 - 2.27 - 263 - 3.09 0 - 0.1 0.3 - 0.1 - 0.8 - 1.1 - 2.7 - 2.1 - 3.9 calc. frequency 21 856.8a 22617.60 21 840.9 20996-8 19572.2 * 8 * * * * * 22498.0a 22907.3 2379 1.1 25439.5 * 'i: * .I: * * 19941.8 24405.6 obs.frequency 21856.8+ 22617*5+ not obs. not obs. 19572.5 22497.9 22906.5 23790.0 25436.8 19939.7 + 24401 *7+ * frequency outside the 17500-26000 Mcls range investigated, f identified by their Stark pattern. b for the method of picking out Q-type transitions compare ref. (5). parameters adjusted to give agreement at these points.DEUTERATED FURANS TABLE 1 (cont.) furan Is.. deuterofuran obs. calc. frequency frequency 19011*46+ 19009.36 2062434 20622054 22540.27 22536.39 2476730 24763.04 * * * a-deuterofuran centrifugal correction 2.10 2.80 3.88 4.46 0.1 0.6 1.3 CalC. obs. frequency fiequency * 8 -k * 24757.0 24757.1 21684.7 21685.3 18212.1 18213.4 a : u'-dideuterofuran obp1.frequency 22696.3" 21 986.4+ 22144.0 21706.9 20961 -2 19942.6 18798.6 1915066 22477.6 22654.8 22998.2 23565.7 24410.9 25575.1 21557:6+ 25628~6~ 18489*2+ 20541.8' 2291 8*6+ 25584.7+ 19666.2 23481.7 25427.9 23934.4 calc. frequency 22696.30 21 986-4a 22144.9 21707.7 20963.4 19945.6 18801.6 * * 19 147.5 22477.7a 226554 22998.5 23565.9 244 10.8 25574.5 21549.8 25620.5 18487.7 20539-9 22915.8 25580.8 19658.9 23472.7 * ;k * * 25436.4 23943.3 centrifugal correction 0 0 - 0.9 - 0.8 - 2.2 - 3.0 - 3.0 3.1 - 0.1 - 0.6 - 0.3 - 0.2 0.1 0.6 7.8 8.1 1 3 1.9 2.8 3.9 7.3 9.0 - 8.5 - 8.9 0 0.1 0 - 0.2 - 0.1 - 2 6 - 3-5 - 5.0 - 0.5 - 0.7 - 1.5 - 1.8 0.1 - 0.9 - 1.4 - 2.1 - 2.2 - 1.7 - 2.0 - 2.6 - 3.3 calc. obs. frequency frequency 21 890.9a 20886.5a 21006.70 20246.7 19095.3 * * 18207.8 2 1027.0 24996.9 21633.2 21965.6 22593.3 23604.0 25063.1 22436.7 * * * 19852.6 22648.5 25830.7 21685.3 * * * 24722.0 22968.5 21 7 10.4 21363.4 21890*9+ 20886*6+ 21006.7 202463 19095.2 18205.2 21023.5 24991 -9 not obs.21965.1 22592.6 23602.5 25061.3 22436.8 19851-7 22647.1 25828.6 21683.1 24720.3 22966-5 21707.8 21360.1 * frequency outside the 17500-26000 Mc/s range investjgated. -1- identified by their Stark pattern. a parameters adjusted to give agreement at these points b for the method of picking out Q-type transitions compare ref. (5). DISCUSSION CALCULATED FREQUENCIES The calculated frequencies were obtained by means of rotational constants which reproduce the two 1 -+ 2 transitions and one of the 3 --f 3 transitions exactly. Since it can be assumed that rotational constants corresponding to rotational states with very low J's come close to the rotational constants for the '' rigid " molecule (undistorted by centrifugal forces) we are justified in labelling the difference between observed and calculated frequency, centrifugal correction.B. BAK, L.HANSEN A N D J . RASTRUP-ANDERSEN 33 The effect is seen to be rather small throughout. In the literature, e.g. StrandbergP has reported centrifugal stretching effects for low J transitions that are very high about 100 Mc/s or more) for molecules like NH3 and H2S. These molecules are, however, of an " open " type, i.e. the distortion of one valence angle can take place fairly independent of the rest of the molecule. Furan cannot be distorted solely by changing the valence angles.Changes in bond lengths have to take place which diminishes the co-ordinate shifts. Whether the rotational constants (given in table 2) derived as explained above are fully correct or not it is worth while noting that the rotational constants for all the isotopic species have been derived by the same procedure so that they are of comparable value. The furan model derived from them at worst corresponds to a not completely rotationless molecule but it may be even better, since to a large extent diflerences between moments of inertia enter the calculations which minimizes the possible error. Our coididence in the approximate correctness of the rotational constants of table 2 may, however, also be strengthened by showing that the centrifugal corrections of table 1, now positive, now negative, are in harmony with a simple physical picture of the rotating molecule.Taking, for example, the three 11 -+ 11 transitions for furan (table l), two of the corresponding centrifugal corrections are negative, one is positive. The general expression for the rotational energy W of a rigid rotor is : Centrifugal distortion changes the rotational constants by amounts AA, AB, and AC and the rotational energy by an amount A W : (2) By differentiation we find (K = (2B - A - C)/(A - C) : W = &(A + C)J(J + 1) + 4 (A - C)E(K). (1) Aw = (3 W ~ A ) A A -t (a w/~B)AB + (3 w p q A c . 3 W/M = +J(J + 1) + +E(K) + +(A - c ) ( ~ E ( ~ ) / ~ A ) ; aE(K)jbf == [2(c - @/(A - c)2](3E(K)/3K) ; 3 W/3B = +(A - C ) ( ~ E ( K ) / ~ B ) ; 3E(K)/bB = [2/(A - c ) ] ( a E ( K ) / a K ) ; a W / ~ C = +J(J + 1) - +E(K) + +(A - C)(JE(K)/JC) ; 3E(K)/>C = [2(B - A)/(A - C)2](3E(K)/aK) ; so that in general AW = (+J(J + 1) + W(K) I- [(C - @/(A - C ) ] ( ~ E ( K ) / ~ K ) ) A A -I- ( ~ E ( ~ K ) / ~ ( K ) A B ( 3 ) For planar molecules coplanari ty is approximately conserved during rotation, so that, since C-1 = B-1 + A-1, AB = (B/c)~Ac - (B/A)2AA.AW then becomes A W 5 [+J(J + 1) + +E(K) + ((C - B)/(A - C) - ( B / A ) ~ } ( ~ E ( K ) / ~ K ) ] A A For furan, A = 9447; B = 9247; C = 4671 ; and K = 0.91614. Therefore, + (+J(J + 1) - +E(K) + [(B - A)/(A - C ) ] ( ~ E ( K ) / ~ K ) ) A C . + [+J(J + 1) - +E(K) -t {(B/C)2 + (B - A)/(A - C))(~E(K)/~K)]AC. A W = [#J(J + 1) + +E(K) - 1.9161 (~E(K)/BK)]AA (4) + [&J(J + 1) - +E(K) f 3.877 ( ~ E ( K ) / ~ K ) ] A ~ .(5) Frequency changes for Q-lines due to centrifugal stretching, are consequently AVQ = vobs. - vcalc., AvQ = [+(E(K)" - E(K)') - 1.9161 ( ( ~ E ( K ) " / ~ K ) - ( ~ E ( K ) ' / ~ K ) ) ] A A - [+(E(K)" - E(K)') - 3*877((3E(~)"/3~) - (~E(K)'/~K))]AC (6) B34 DEUTERATED FURANS where double-prime quantities belong to the upper rotational level. The ap- proximate validity of (5) can only be maintained for cases in which AA (and AC) are almost the same for the two states in question. This condition is fulfilled for the three 11 -+ 11 transitions here considered, i.e. 119.3 -+ 119.2 ; 118.3 -+ 1110.2 ; and 119.2 --+ 1111.1. For all five states involved J = 11. The approximate magnitude of J‘s component on the approximate symmetric-top axis (perpendicular to the plane of the molecule) is 1, 2 and 3 so that 5.is oriented at small angles to the molecular plane. AA and AC, therefore, mean some sort of average corrections to Afigid and Brigid, valid for J = 11 and K+1 = 1, 2 and 3. Values of E(K) and ~ E ( K ) / ~ K are taken from Turner’s tables.7 For the three 11 -+ 11 transitions we get AVa(I19.3 -+ 119.2) = - 4.98 = - 44.783 AA + 95.040 AC, A ~ ~ ( l l 8 . 3 -+ 1110.2) = - 2.63 = 31.384AA - 58.170 AC, AV~(l19.2 + 1111.1) = 4.46 = 133.603 AA - 265.033 AC. No rigorous solutions exist since the equations are only approximately correct. But the values AA = - 1.385 Mc/s, AC = - 0.713 Mc/s give AVQ = (in the order above) - 5-73 ; - 2-00 ; 3.92 Mc/s with correct sign and correct order of magnitude (compare table 1).The calculation may be checked by repeating it for the three 8 -+ 8 transitions, 86.3 -+ 86.2; 85.3 -+ 87.2 and 86.2 --f 88.1. Ap- propriate values of b l and AC may again be found, and we can furthermore predict that AA for the J = 8 levels here involved should be approximately - 1.385 X (8/11)2 = - 0.73, while AC (J = 8) - - 0.713 x (8/11)2 = - 0.377. By insertion we actually find AA = - 0.62 and AC = - 0.33. The calculated centrifugal corrections are - 2-41 ; - 1.26 ; and 1.65 Mc/s (compare table 1). The magnitude of the centrifugal effect on the two 1 --f 2 transitions so important for this calculation could now be estimated by means of (5). Approximately AA(J= 2)/AA(J= 1) -4. Also, &(J= 2) must be about 25 times smaller than AA(J = 11).We then find that the two 1 -+ 2 lines could be in error by 0.2 -+ 0.3 Mc/s due to centrifugal stretching. To find approximately correct values of (A - C ) and K we had initially made the usual plots 5 of AE(K)/V~~~. against K. All the points in the thick swarm of points of intersection had been determined and an average ( A - C ) and K had been taken. It turns out that these values only deviate insignificantly from the rotational constants based on the two 1 -+ 2 transitions and one of the 3 -+ 3 lines (see table 2). We can now see why. Since for a planar molecule AK = @/(A - C))[- (B/A)2AA + (@(A + B))/A2)ACI, we calculate AK = - 0.00005 for the J = 11 levels, which is confirmed by looking at the graph. For the J = 8 levels it is about - 0.00004.Since levels from J = 3 to J = 12 are involved in the graphical method we see that K found by this pro- cedure would not be likely to deviate much from the true value. Also, the value of ( A - C) found graphically was astonishingly correct. We can see now that A(A - C) - - 0.7 Mc/s for the J = 11 levels and - 0.3 Mc/s for the J = 8 levels. Again, only a small deviation from the rigid rotor values is to be expected. It is interesting to note that, at least for furan, the influence of the zero-point fluctuations is of comparable magnitude to the effect of centrifugal distortion. In the calculations to follow we are forced to assume that, e.g. the C-D distance in or-deuterofuran is identical with the corresponding C-H distance in furan.However, we know from other examples (methyl halides) that the hydrogen x-co-ordinate (fig. 1) may be about 0402A greater than for deuterium. In equalizing these two co-ordinates we are ignoring in 1 ~ . Since x = 2.047 A and 1~ - 54.6 a.m.u.& the percentage error in I’ is 0.014. The percentage error in A is the same, so that AA = 0.76 Mc/s, i.e. rnH[(X + 0.002)2 - x2] - rn,x(0.004)B . BAK, L. HANSEN A N D J . RASTRUP-ANDERSEN 35 close to the effect of centrifugal stretching. The consideration shows why one should be careful not to overload a molecule with deuterium atoms since the effect of zero-point energy increases to a not unimportant extent. RESULTING ROTATIONAL CONSTANTS In table 2 we have summarized the calculated values of K, rotational constants, etc.derived (column 1) from the two 1 -+ 2 lines and one of the 3 --+ 3 lines, and by taking ( A - C)/2 and K from the Q-line plots and finding ( A + C)/2 in connection with the two 1 -+ 2 lines (column 2). TABLE 2.-cALCULATED VALUES OF ROTATIONAL CONSTANTS (A, B, c) MC/S, ( A - c)/2, ASYMMETRY PARAMETER ( K ) AND PRINCIPAL MOMENTS OF INERTIA (ZA, IB, IC) a.m.u. w2 OF FURAN AND DEUTERATED FURANS TOGETHER WITH THE QUANTUM DEFECT (q.d.) FOR Q-LINE PLOTS, ETC. (see text) ALL FOUR SPECIES. COLUMN (I), FROM THE THREE LOW-J LINES; COLUMN (2), FROM (1) A 9447.04 B 9246.77 C 467044 ( A - C)/2 2388.10 K 0.9 1 6 14 ZA 53.5067 IB 5 4 - 6 6 5 6 IC 108.22Ofl q.d. 0,048 1 furan (2) 9446.98 9246.72 4670.86 2388-06 0.91614 53-5070 54.6659 108.2199 0.0470 a-deuterofuran (1) - (2) 0.06 0.05 - 0-02 0.04 0*00000 - 04003 - 0.0003 0.0005 0.0011 (1) 9280.1 5 8638.74 4472.05 2404.05 0.73320 54.4689 58.5132 113-0309 0-0488 (2) 9280.15 8638.56 4472.09 2404.03 0.73312 544689 58.5144 1 13.0299 0.0466 (1) (2) 0.00 0.18 - 0.04 0.02 0~00008 o*oooo - 0.0012 0.0010 0.0022 B-deuterofuran a : a'-dideuterofuran A 9383.77 B 8490.39 C 4455.47 ( A - C)/2 2464.1 5 K 0.63745 I A 53.8675 ZB 59.5355 ZC 113.4516 q.d.0.0486 938355 8490.36 445549 2464.03 0.63751 53.8687 59.5358 1 13.4510 0.0465 0.22 0.03 - 0.02 0-12 - 0.00006 - 0.0012 - 0.0003 0.0006 0.0021 9033950 8 160.64 428 5.80 2373-85 0.63230 55.9562 61.9412 117.9430 0.0456 9033.38 8 160.68 4285.82 2373.78 0.63236 55.9569 61.9409 1 17.9424 0.0446 012 - 0.04 - 0.02 0.07 - 0*00006 - 0.0007 0.0003 0.0006 0~0010 RESULTING MOLECULAR MODELS In the calculation of the molecular models compatible with these constants we started to find the co-ordinates of the hydrogen atoms in the system shown in fig.1. By insertion in the formulae given in our paper on pyridine,s we get the results of table 3, based on the principal moments of inertia for furan, a- and P-deuterofuran. TABLE 3.-cALCULATED POSITIONS OF THE HYDROGEN ATOMS IN FURAN ( X , y ) ; MOMENTS OF INERTIA OF THE HYDROGENS (IH) AND THE CARBON-OXYGEN RING (14c.0) from rotational constants (1). u-hydrogen fi-hy drogen a-hydrogen j3-hydrogen 2.048 1.376 2.047 1.377 4.192 1.894 4.1 89 1.896 - 0.811 1.839 - 0.813 1.838 0,657 3.380 0.660 3.378 8.14 8.14 46.53 46.53 from rotational constants (2). 12.27 41.24 12.27 41.2436 DEUTERATED FURANS These differences are so small that they can be ignored in view of the error and the experimental uncertainty in the rotational constants.5 From the co- ordinates of the hydrogens and the rotational constant B for a : a'-dideuterofuran ' c l A , FIG.1 .-Co-ordinate system used through- out. The x, y and z axes are the principal inertia axes for furan. The origin is the centre of mass, one calculates I;'*" = 41.23 which agrees excellently with I;'* " from table 3. By properly taking into account the difference in the position of the centre of mass when comparing furan with a : a'-dideuterofuran, one calculates 1x4's " = 46.52 which also fits with the value given in table 3. We shall only proceed with the co-ordinates and corresponding moments of inertia given under (1) in table 3.Following the same line of thought as in our paper on pyridine 5 we finally arrive at the models given in table 4. VALENCE THEORY ASPECTS In the last three years furan has twice been subject to extensive theoretical treatments 8 9 9 in which its resonance energy, dipole moment, ultra-violet ab- sorption, etc., have been calculated. Both papers build upon the electron- diffraction data by Beach,2 where d(C-C) = 1.46 & 0.03 A, d(C=C) = 1.35 A (assumed), and d(C-0) = 1.40 f 0-03 A. Whichever of the models (I)-(V) above one chooses it is seen that a renewed treatment is highly justified and we intend to carry this out. Without these detailed calculations, however, quite a few important features may be stated. TABLE 4.-FURAN MODELS (1-v), COMPATIBLE WITH THE MICROWAVE SPECTRA OF FURAN, GIVEN AT THE TOP OF THE TABLE a- AND P-MONODEUTEROFURAN, AND a : a'-DIDEUTEROFURAN ASSUMING THE C-H DISTANCES I I1 V I11 IV 1 -070 1 a070 1.356 1.370 1 -444 107" 30' 110" 29' 105" 46' 116" 41' 127" 45' 1.080 1.080 1.387 1.337 1.438 104" 24' 111" 26' 106" 22' 113" 54' 127" 33' 1.075 1.075 1.371 1.354 1 a440 106" 00' 110" 55' 106" 05' 115" 20' 127" 40' 1 -070 1,080 1.374 1.357 1.426 106" 24' 110" 14' 106" 34' 115" 28' 127" 56' 1.080 1 -070 1.367 1.350 1.458 105" 38' 111" 43' 105" 28' 115" 18' 127" 17' First, the choice of models (I)-(V) seems reasonably representative. It is seen that consideration of C-H distances shorter than 1.07A would result in a C-0 distance less than 1-356 A.Since the normal C-F distance is 1.385 A, (CH3F10), such a small C-0 distance could only be accepted very reluctantly. Also, taking C-H distances longer than 1-08A would mean that the C=C distance would decrease still further below 1.353 A, the " unperturbed " C=C distance.11 Since all the models agree in having the C-C distance rather short (- 1.44A) it follows from well-established bond-order, bond-distance curves that about 24eB.BAK, L. HANSEN AND J . TASTRUP-ANDERSEN 37 must be present in the C-C “single” bond. The charge +e could have come from (i) the two C=C bonds, (ii) the oxygen atom, but not from the C-0 bondi which are so abnormally short (the C-0 distance 12 in CH30H is 1.434A). Considerations concerning the dipole moment p to be presented later show that the alternative (i) must be the more correct.Therefore, each of the C=C bonds contains only about 32e which means that they must be slightly longer than a normal C=C bond. Thus, models like (I) and (11) can be rejected. It is hard to dis- tinguish between models (III), (IV) and (V), although (V) may be slightly less probable because of its short C=C distance. The models (III) and (IV) essentially agree in all interatomic distances and their correctness may be verified by con- sidering the p of furan. Since the C-0 distance in these models is close to 1.372 A there must be about 22e in each of the C-0 bonds. If the “extra” ge is taken from oxygen this atom gets the charge + 3e. Fig. 2 shows the approximate distribution of what may be termed the redistributed n-electron charges in concordance with the above considerations.While tetrahydrofuran has p = 1.68 D,13 the electric moment of furan 1 is only 0.661 D. Furan may now be thought of as having (i) a pa equal to that of tetrahydro- furan (so-called a-moment), presumably directed downwards in fig. 2 (drawing the vector from + to - charge), (ii) a super- imposed so-called pmig, the moment due to the migration of the n-electrons. The m--charges in the three C-C bonds have been placed in the middle of the bonds and a simple calculation shows that this part of b i g approximately outbalances the 0-moment. A resulting moment of 0.66 D may now easily be explained by means of the charges on the oxygen and in the two C C C C FIG. 2.-Migrated n-electron charges in furan as derived from bond length considerations.C-0 bonds (where we need not locate the charge in the middle). A necessary consequence is, however, that the resulting p-vector points upwards in fig. 2. Confirmation of this result may be found in that the p of 2-methylfuran14 is larger than the p of furan. Its p is 0.74D. Models like (III) and (IV), therefore, explain the microwave spectra of 4 iso- topic furan species and they seem to fit reasonably well with the measured dipole moment. It may be added that a renewed electron-diffraction investigation 15 (rotating-sector method) has shown that d(C-H), -- 1.074-1-077 A while which fits excellently with our models. The older model of Beach 2 from electron- scattering has this average equal to 1.392A. It remains to show by means of thorough calculations how the models may be fitted to the observed ultra-violet spectra 16 and with current ideas on the resonance energy, etc., of furan. (2d(C-0) + 2d(C= C) + d(C-C))/S = 1.377 A 1 Sirvetz, J. Chem. Physics, 1951, 19, 1609. 2 Beach, J. Chem. Physics, 1941, 9, 54. 3 Gilman and Wright, J. Amer. Chem. SOC., 1933,55, 3302. 4 Clauson-Kaas, Nedenskov, Bak, Rastmp-Andersen, Acta Chem. Scand., 1954, 8, 5 Bak, Hansen and Rastrup-Andersen, J. Chem. Physics, 1954,22,2013. 6 Strandberg, Ann. N. Y. Acad. S., 1952, 55, 808. 7 Turner, Hicks and Reitwiesner, J. Chem. Physics, 1953, 21, 564. 8 Simonetta, J. Chim. Phys., 1952, 49, 68. 1088.38 METHYL DIACETYLENE 9 Nagakura and Hosoya, Bull. Chem. SOC. Japan, 1952,25, 179. 10 Gilliam, Edwards and Gordy, Physic. Rev., 1949, 75, 1014. 11 Gallaway and Barker, J. Chem. Physics, 1942, 10, 88. 12 Ivash and Dennison, J. Chem. Physics, 1953, 21, 1804. 13 de Vries Robles, Rec. trav. chim., 1939, 58, 111. 14Nazarova and Syrkin, Izvest. Akad. Nauk. S.S.S.R. Otdel. Khim. Nauk., 1949, 35; 15 Almenningen, Bastiansen, and Hansen, to be published. 16 e.g. Mackinney and Temmer, J. Amer. Chem. SOC., 1948, 70, 3586. Chem. Abstr., 1949, 43, 4913.