An Electron Diffraction Study of Ketene Dimer, Methylketene Dimer

Giovanni Morales and Ramiro Martínez. The Journal of Physical Chemistry A ... Giovanni Morales, Ramiro Martínez, and Tom Ziegler. The Journal of Phy...
1 downloads 0 Views 1MB Size
April 5, 1955

DIFFRACTION O F KETENE DIMER,METHYLKETENE DIMERA N D TABLE V Relative methyl affinities

Compound

1 Benzene Diphenyl ether 2.5 3 Pyridine 5 Diphenyl 11 Benzophenone 22 Naphthalene 27 Phenanthrene 29 Quinoline 183 (trans) Stilbene 57.5 Chrysene 125 Pyrene 468 Benzanthracene 820 Anthracene D. S. McClure, J . Chem. Phys., Nauman, Thesis, Berkeley.

E s - T , cm.-l

Ref.

29,400 28,200 26,050 22,800 24,100 21,300 21,600 21,700 21,700 (cis) 19,800 16,800 16,500 14.700 , - -

a, b

~~

17, 905 (1949).

b b b a a, b a a

b U U

a a

* R. V.

fers to the trans-isomers while the phosphorescence was measured for the cis-isomer. Consequently, the discrepancy may result from the difference in

[CONTRIBUTION FROM

P-PROPIOLACTONE

1955

All these results the properties of the show definitely that the suggested model for the reaction is plausible, and the assumptions introduced are justified a t least in the first approximation. The above discussed correlation implies that one might predict the reactivity of an aromatic molecule toward radicals from the observed singlet-triplet excitation energy, or vice versa. Finally, we wish to mention that this type of investigation can be extended to many classes of compounds, and indeed the above method has been applied recently in determining the reactivities of methyl radicals toward vinyl monomers, quinones, substituted aromatics and other compounds. These results will be published later. In conclusion, we would like to thank the National Science Foundation for financial support of this investigation. (23) Indeed i t is reported t h a t the c i s isomer is much less reactive than the trans, see, e.g., Marvel and Anderson, THISJ O U R N A L , 76,5434 (1954). SYRACUSE,

x. Y .

DEPARTMENT OF CHEMISTRY,

CORNELL UNIVERSITY]

An Electron Diffraction Study of Ketene Dimer, Methylketene Dimer and P-Propiolactonel B Y JUDITH

BREGMAN AND

s. H. BAUER

RECEIVED NOVEMBER 4, 1954 The results of electron diffraction studies of the structures of P-propiolactone, and the dimers of ketene, and monomethyl ketene in the gas phase are reported. Procedures for the reduction of diffraction intensity data obtained with a rotating sector are described. A comparison is made of the structural parameters found with those in related straight-chain and small ring molecules.

The structure of diketene has been of interest due to its great reactivity. This reactivity has meant, however, that by following standard organic methods it has not been possible to reject any of the following five configurations conclusively.2 U

I

H2C-C

I

II

HC-C

I

I I1 diketone mono-enol of I

I11 IV V vinylaceto- P-crotono- acetylketene &lactone lactone

(1) A preliminary report on the structure of diketene was presented a t the American Chemical Society Meeting in April 1946, “An Electron Diffraction Study of Ketene and Dimethylketene Dimers” by S. H. Bauer, J. Bregman and F. W. Wrightson. Some of this material was presented a t the American Society for X-Ray and Electron Diffraction meeting in June 1949. “An Electron Diffraction Investigation of the Structures of Beta-propiolactone and Ketene Dimers,’’ Judith Bregman and S. H. Bauer, and some a t the American Crystallographic Association meeting in April 1950, “The Reduction of Electron Diffraction Photographs and the Computation of Radial Distributions from Scattering Data,” K. P. Coffin, Judith Bregman and S. H. Bauer. (2) P. F. Gross, Thesis, Cornell University, 1936; A. B. Boise, Jr., I n d . Eng. Chem., 32, 16 (1940); C. D. Hurd and J. L. Abernethy, THISJOURNAL, 62, 1147 (1940); F. 0 Rice and R. Roberts, i b i d . , 65, 1677 (1943): J. T. Fitzpatrick, i b i d . , 69, 2236 (1947); A. T. Blomquist a n d F . H. Baldwin, i b i d . , 70, 29 (1948); C. D. Hurd and C. A. Blanchard, i b i d . . 72. 1461 (1950).

Measurements of the dipole moment and other additive physical properties eliminated only the symmetric configuration I. The absorption spectra of the liquid phase4 have been interpreted as consistent with one or another of the lactone configurations, I11 and IV, or possibly requiring a mixture of them. Recent studies of the exchange of H by D on treating diketene with CH30D5support the presence of I11 alone in the liquid phase. Mass spectrograph studies6 indicate that the configuration in the vapor is predominantly, if not entirely, 111. A determination of the crystal structure by X-ray diffraction’ also reported the configuration to be 111; the angles and distances agree with those (3) F. Chik and N. T. M . Wilsmore, J . Chem. S o c . , 93, 946 (1908); 97, 1978 (1910); %’. R. Angus, A. H . Leckie, G. L. LeFevre, R. J . W. LeFevre and A. Wasserman, i b i d . , 1751 (1935); C. D. Hurd and J. W. Williams, THIS JOURNAL, 68, 962 (1936); P. F . Oesper and C. P. Smyth, {bid.,64, 768 (1942); E. C. Hurdis and C. P. Smyth, ibid., 65, 89 (1943); J . D. Roberts, R . Armstrong, R. F. Trimble, J r . , and M . Burg, i b i d . , 71, 843 (1949). (4) G . C. Lardy, J . chim. p h y s , 21, 281, 333 (1924); M. Calvin, T . T. Magel and C. D. Hurd, THIS JOURNAL, 63, 2174 (1941); H . J . Taufen and M. J. Murray, i b i d . , 67, 754 (1945); D. H. Whiffen and H. W. Thompson, J . Chem. S o c . , 1005 (1946); F. A. Miller and S. D. Koch, Jr., THIS J O U R N A L , 7 0 , 1890 (1948); R . C. Lord, Jr., R. S. McDonald and R . J . Slowinski, unpublished work, see J. D. Roberts, et a1 , r e f . 3. (5) J. R Johnson and V. I. Shiner, Jr., THIS J O U R N A L , 75, 1350 (1953). (6) F. 4 . Long and Lewis Friedman, i b i d . , 75, 2837 (1953). (7) W. Lipscomb and L . Katz, Acta Cryst., 5 , 313 (1952).

1956

Vol. 77 TABLE I Sample temp., Compound

OC.

M.P., "C. lit.

Dimethyl ketene dimer

100

8-Propiolactoiie Diketene Methyl ketene dimer

a

M.P., O C . before run

After run

114-116

114-115

100

-33.4

-33

113-115 114-115 -32

60-75 75-90 60-75

....

J. D. Roberts, et al., ref. 3.

-7

* C. M . Hill, ref. 8.

- 7

- 7

-52 - 11

- 52

Obsd. n% Lit.

1.4110 (240) 1,4368 1.4322" 1.4290* 1.4280"

IJefore run

After run

1.4108 1.4109 1.43G9 1.4351

1.4108 1.4367 1.4351

J. C. Sauer, ref. 8.

found in this investigation and will be discussed later. The electron diffraction study was begun when only the dipole and some of the spectroscopic evidence were available. At least two products of direct dimerization of iiiethylketene have been reported,8 a solid (with m.p. 140") and a liquid (b.p. 50-52" a t 9 mm.). An acidic dimer, which is generally assumed to be identical with the solid, can be prepared from dimethyl a,a'-dimethylacetone carboxylate. e Woodwardlo has shown this compound to be the monoenol of the cyclic P-diketone (configuration of type 11, but with two hydrogen atoms not on the same carbon atom replaced by methyl groups). The liquid dimer, although easily prepared by the same type of reaction as that yielding the dimer of dimethylketene, has properties similar to diketene and markedly different from those of dimethylketene dimer. I t was this dimer that was studied by electron diffraction. The chemical evidences permits consideration of five configurations homologous to those for diketene. Configurations of type I, 11, and V can be eliminated as in the case of diketene: I by the large dipole moment, I1 and V by the Raman and infrared spectra which were found consistent with either lactone and may require a mixture of then^.^^^ The configuration of P-propiolactone is known from chemical evidence." I t was included in this investigation to provide diffraction patterns for comparison with those of diketene. In addition, the values of the parameters in the four-membered strained ring are of interest. Dimethylketene dimer is known to have a configuration I , excluded for ketene dimer, and had previously been studied by electron diffraction.I2 Photographs of this compound were taken for comparison with those of diketene and to aid in developing a method for handling microphotometer traces of sector pictures. Experimental 7'he earliest diffraction patterns of diketene were obtaiiied without the use of a sector.13 From these i t was apparent (8) C . RI. Ilill, Thesis, Cornell University, 1941; J . C. Sauer, THIS 69, 2-114 (1947); H. Staudinger a n d H. W. Klever, Ber., 44, 533 (1911). (9) G. Schroeter a n d C. Stassen, i b i d ,40, 1604 (1907). (10) R . B. Woodward and Gilbert Small, Jr., THISJOURNAL, 72, 1297 (1950). (11) T L. Gresham, L. E. Jansen, F. W . Shaver, e # al , i b i d . , 70, 9913 a n d Fi (19.18) ( 1 2 ) \V N l,i[m%mh and V . S r h # m a k e r , J . Chem P h y r , 14, 47.:

JOURNAL,

(I!) Lli)

(13) The wurk of Dr Frances AI. Wrightson, who obtained the first diffraction patterns and made the approximate calculations, is gratefully acknowledged.

that the diffraction pattern contained complicated features whose estimation by the visual method was not sufficiently reliable for quantitative comparisons. Therefore, sector pictures, from which a more objective estimate of the diffraction pattern may be obtained, were taken of all four compounds. Samples of the three lower molecular weight materials were vacuum fractionated and stored in vacua a t Dry Ice temperatures. Dimethylketene dimer was vacuum sublimed in the presence of anhydrite. Portions used for the electron diffraction studies were transferred under high vacuum to holders which fitted onto the camera. Physical constants obtained on these samples are given in Table I . Measurements after runs were made on material sublimed from the residues in the sample holders. The methylketene dimer was not pure, as evidenced by the two melting points of one cut and a decrease of index of refraction in successive cuts in the vacuum fractionation (from 1.4359 t o 1.4341). It seems probable that the impurity was propionic anhydride (see below). It is interesting to note that the refractive indices reported in previous studies, as well as those of possible impurities resulting from the preparation used,]4 are all lower than those observed in this invest igation The electron diffraction photographs were obtained with the camera and high temperature nozzle described by Hastings and Bauer.15 The camera conditions were as follows

.

h'ozzle plate distance, cm.

Sector Kon-sector

Wave length,

A.

Iiiltn

Max s obtained

19.12 0,0691 Microfile 12-14 19.20 ,0539 Eastman comnicrcial 27-30

Analysis of the Data.-Visual estimates of the intensity and position of the maxima and minima were made from both sector and non-sector pictures. Microphotometer traces of the sector pictures were also obtained and treated as described below. The usual comparison of the observed intensity curves with computed curves for various models were made, both with the visual data and the reduced microphotometer trace. These intensity curves were computed according t o the eauation

(where the symbols have their usual tncanings) to be coinparable with the sector data. Most of the calcu1:ttioiis with variable coefficients were made with I.B.M. machiries, using the method of progressive digiting.I6 The microphotometer data allowed a quantitative comparison of computed and observed relative intensities. This comparison, the "specific contrast," defined as [ Z ( S ) / I I , ~ ] ~ ~ ~ / [~(s)/Zb~]~~l~ and computed a t the observed maxima anti minima, provides another criterion by which postulated models may be judged. Two radial distribution curves were calculated for each of the four compounds, from the visual data, and from the (14) T h e dimer was prepared from p r w i o n y l chloride fm p. -94', 1.4051) and triethylamine (m.p. -11-1.8", n% 14003). blethylketene formed and dimmized. Had moisture been present, propionic acid (m.p. - 2 2 ' . n% 13874) a n d p r o p i < > n i c.tnhydride (111 Ii n% 1.4038) might also have formed (16) J. M. Hastings and S . H. Bauer, 3. C h e m . P h y s . , 18, 13 (1950) (16) This procedure and the relntively sinall card file needed will be described in a forthcoming puhlicatiim ( J H ) 11%

OF KETENE DIMER,METHYLKETENE DIMERAND 6-PROPIOLACTONE April 5, 1955 DIFFRACTION

1957

microphotometered sector data extended with visual data from noli-sector photographs. The calculations were made using punched card machines according t o the equation

rD2 (r) =

5

qie-bzg?

i(pi) sin

(6 qir)

A

pi = 1

where s = ~ q i / l O ,ba was chosen so that e-b’a: E 0.1 a t qi = qmax, n g 100, and i ( q i ) is directly proportional t o the observed intensities, except for small pi where the intensity curves were “drawn-in’’ from preliminary calculations of

4s).

The microphotometer traces of the sector photographs must be corrected to take out extraneous contributions to the scattering, and for variation of the atomic scattering function with angle, before they can be used in calculations as a measure of the molecular scattering. This process, drawing in a background on the microphotometer trace, itself yields structural information. The background can be positioned using the theoretical intensity curves as follows: the microphotometer densities a t those s values for which the molecular contribution t o the scattering is zero (the “cross-over points” where the atomic and molecular curves intersect) would lie on a straight line if the diffraction experiment were ideally conducted. Each theoretical model provides a set of such points which can be marked on the photometer record. Even if no set lies on a straight line, those sets of points given by fairly satisfactory models scatter about a line whose curvature changes slowly compared t o the molecular oscillations. This curve may be considered t o be a “reasonable background.” In addition, the better the model, the closer the cross-over points lie t o the background. Thus variations in parameters can be followed quantitatively. The positions of crossover points on the microphotometer trace for two configurations of diketene are shown in Fig. 1. The points from intensity curves for models with the lactone configuration, curve A, lie much closer t o a reasonable background than do points from models based on a straight-chain configuration, curve B. A similar procedure is possible using the maxima and minima of the photometer trace. The heights of t h e maxima and minima of the computed intensity curves can be reduced so that they all have the same value a t oiie peak. Then for each model, a t the positions of the computed maxima, multiples of these reduced heights can be marked on the graph of the photometer data, measuring downward from the microphotometer trace. Correspondingly, a t the positions of the computed minima, multiples of the reduced depth can be marked on the graph, measuring upward from the trace. For accurate data and the correct structures, the points corresponding to some multiple should lie on a straight line. Experimentally they do not. However, the requirements that the background should curve only slowly and that the multiples which fall on it should change slowly with s, allow a “reasonable background” to be drawn. And the better the model, the more nearly will this background lie on points of a constant multiple. Variation in this multiple measures the same effect as variation in the specific contrast. Curves C and D of Fig. 1, show cross-over points and maxima-minima multiples for two models, marked on the microphotometer trace of the dimethylketene dimer sector photographs.

Results Dimethylketene Dimer.-Our results are in good agreement with those of Lipscomb and Schomaker,12 but have one interesting new feature. A high-resolution microphotometer trace of the sector patterns showed the ring near s = 4 as a maximum, although OUT visual interpretation and the Leeds and Northrup microphotometer both indicated a shoulder (Fig. 2). The radial distribution curve analysis (Fig. 3) follows Lipscomb and Schomaker’s. The point of interest for comparison with the diketene results is the small peak a t r = 2.19 8. which corresDonds to the two diaconals in a four carbon ring (2.20 A.) plus the twclve non-bonded C-H (2.17 A.).

B

\

I

V n

7

14

S+

Fig. 1.-Microphotometer trace background analysis: A and B, cross-over points from intensity curves for models based on two configurations of diketene; C and D, crossover and maxima-minima points from intensity curves for dimethylketene dimer, model A (Lipscomb and Schomaker) : C-C = 1.56 A., C-CH, = 1.54A., C=O = 1.22A.; LCCO-C = QO”, LCHa-C-CHa = 109’ 28’; a i j * (bonded) = 0.0015, aij2(unbonded) = 0.0022 except: C, aij2 = 0.0061 for heavy atom distances above 3 A.; aijZ = 0.0043 for hydrogen distances above 2.2 A. D, all hydrogen distances above 2.2 A. omitted.

Correlation of the visual and microphotometer intensity curves with calculated curves yields new results only as regards details affecting the intensity curves near s = 4. Lipscomb and Schomaker omitted the long hydrogen distances (above 3 8.) in their calculations. Inclusion of these terms converts the shoulder on the computed curves a t s = 4.5 to a maximum. The need for these hydrogen distances is also indicated by the microphotometer trace, both from the calculation of contrast (Table 11), and from the background analysis; compare TABLE I1 DIMETHYLKETENE DIMER: MICROPHOTOMETER DATA S

Max.

Min.

RelaDev. tive from unit height, background, visual mpr.

1.7 2.7 3.6

6.0 7.0

8.0 9.0 10.0

11.0 12.9

Av. s = 3.6-11.0 Av. dev.

25 30 35 40 25 20 14 14 20 14

-0.180 ,064 - .204 .128 ,076 .070 - .036 .028 .038 .046

-

Specific contrast mpr./D mpr./C

2.00 . ~for both the visual and photometer data are given in Table IV for model B. The structural information obtained from drawing a background onto the microphotometer trace is consistent with, though not as complete as that from the correlation and radial distribution analyses. All the chain models may be rejected. All the unsymmetric rings are acceptable; for the vinyllactone rings the background passes through maxima-minima points of constant multiple. However, the differences among the rings are smaller than the differences between them and the "reasonable" background position finally chosen.

I

TABLE IV DIKETENE:INTENSITY CURVEDATA Max.

Min.

Visual

0

2.24 2.88 3.83 4.45 5.13 6.20 7.46 8.53 9.43 10.36 11.47 13.26 13.86 14.31 16.22 18.28 20.47 22.51 23.52 24.41 26.49

1 1 2 2 3

3 4 4 5 5 6 6 7 8 9 9 10 11

11 12 Av. Av. dev.

1 3 3 5 5

so

Mpr.

B/vis.

4.01

0.935"

SC/SQ

B/mpr.

0.893" I 08

6.30 7.80

10.18 11.87

0.960" 1.013 1.028 0.972 0.9% 1.041 0.995

0.944' 0.969

100

c 1.08t

0.998 1.007

0.965 0.993

1.005 1,018 1.002 1 .00 0.02

--N

I

0.99 0.015

Relative height, visual

Dev. from unit background, mpr.

Specific contrast , mpr./B

35 35 26 17.5 21

-0.108 .IO4 - .030 .030 - .020

0.50 .62 .39

Av. Av. dev. a Omitted in computation of average.

I 10

I

5

.68

.63 .56 .I0

Electron diffraction results lead to the conclusion that diketene in the vapor state has the vinyllactone configuration, 111, although the possibility of a mixture of I11 with a small amount of the crotonolactone configuration, IV, cannot be excluded by this method. Assuming no mixture, the satisfactory model has distances and angles similar t o those found in the /3-propiolactone; only the C-O distance differs appreciably. An acceptable set of parameters ar: C=O = 1.19 A,, C=C = 1.31 A., C-0 = 1.41 A., C-C = 1.52 A., LC-0-C = 89'.

I

I

I5s

25

2o

Fig. 5.-Diketene, models investigated. Parameters for all curves unless otherwise specified: Rings planar. H C H plane I to ring. Other groups attached t o the ring lie in its plane and bisect the exterior angle. C-C = 1.54 A., C-0 = 1.42 A., C-H = 1.09 A., C=C = 1.33 A., C=O = 1.21 A., 0-H = 0.97 A. L H C H = 109' 28' in -CH3 groups, L H C H = 120' in C H I groups. A, C-0 = 1.47 A., L C O C = 89'; B, L C O C = 89'; C , HzC-CO = 1.56 A., C-0 = 1.40 A., C=C = 1.36 A,, C=O = 1.23 A., LCOC = 97'; D, L C O C = 97'; E, L C O C = 90'; F , L C C C = 86"; G, L C C C = 90"; H , asdrawn(cis), with LH,CC-C = llOo, LO=C-C = 125", L C C H = 120°, LCC=C = 120"; I, CHaCO rotated through 180' from position in H (trans); J, angles as in H . OC-C = 1.48 A,, C=C = 1.37 A., O=CCH, = 1.28 A., O==CCH = 1.24 A,; K, trans to J ; L, (cis) LHvC-C-C = 110'. LO=C-C = 130°, LC-C=C = 130°, L C C H = 115'; M, CHiCO group rotated 90' from position in H ; N, equal weights of H+I+M.

HzC=C-CHz

A, B,

I / 0-c=o

H3C-C=CH E,

I t 0-c=o

1962

+

BREGMAN AND S, H . BAUER

JUDITH

I

5

I

I

I0

15 5.

I 20

1 25

I

30

Fig. 6.-Methylketene dimer, models investigated. The parameters for all curves unless otherwise specified are: rings planar. CHI-C(ring)-H plane I to ring. Other side-chains coplanar with the ring and bisect the exterior angle. C-C = 1.54 A., C-0 = !.42 A,, C=C = 1.33 A,, C=O = 1.21 A., C-H = 1.09 A. LHCH- 109' 28' in -CH, groups, LCH3-C-H = 109' 28' adjacent to ring, = 120' adjacent to C=C. Both -CH3 LCH3-C-H groups are not attached to the same carbon. B, as drawn. L C O S = 97", C-0 = 1.40A., C=C = 1.36A., C=O = 1.23 A . ; A, like B except 180' rotation around C=C; C, LCCC = go', CHJ groups cis to each other; D, as = drawn except 60" rotation of CH&Hg. Le-CO-C llOo, LO=C-C 125", LC-C=C = 120°, LOC-CCCH3 = 120', LOC-CH2-CH3 = 109' 28'; E, like D except 90' rotation of CHaCCO; G, like D except 180' rotation of CHaCCO; F, like G except 120" rotation of C H J C H ~so both CHa groups coplanar with chain; H, free rotation: E G. equal weights of D

+ +

A, B

D. E

C CHs

H CHJ HJCC=C-C-H I

1

O-b=O

O=C-C-H I

1

H3Cb-C=0

CH3 HZ I H3CCCC=C=O II

0

The side-chain angles were not investigated in detail. Those found by Lipscomb and Katz in the = 145") LC=C-C = 136", crystal, LC-C=O

VOl. 77

are in satisfactory agreement with the electron diffraction data. The limits of error, excluding simultaneous variation of parameters, are 0.04 8.and 4". If this restriction is removed, the limits must be larger, although considerations similar to those given earlier again apply. Methylketene Dimer.-Two types of diffraction patterns, differing in the range s = 4-19, were apparent in the non-sector pictures of methylOnly one type, b, was ketene dimer (Fig. 6). found on the sector photographs. The effect may be due to the impurity (possibly propionic anhydride) believed to have been present. It was found after completing the photographs that the beam shutter was defective. Since i t was not always open when intended, it may have allowed the scattering due to one compound to be selectively recorded in each picture, due to a difference in volatility of the two. Both patterns were used to compute radial distribution curves, Fig. 3 . Pattern a yielded a curye with peaks a t r = 1.14, 1.53, 2.29, 3.28 and 3.90 A., which cannot be interpreted as representing the distances in any of the possible methylketene dimer configurations. The C=C and C-0 distances are missing; the 2.29 8. peak is unexplained. This curve is consistent with an acid or acid anhydride structure which has no C=C bonds: the 0-0distance is expected to be near 2.30 A. Propionic anhydride, a possible impurity in the preparation used, melts a t -Go,and its presence would account for the lower of the two observed melting points. The analysis of the distribution curve from pattern b is straightforward in terms of some of the possible configur$ions for methylketene dimer. The peak a t 1.10 A. corresponds to the primary disIt occurs a t a smaller valtances C-H and C=O. ue of r than expected, poossibly due to the large oscillationq below r = 1 A. The next maximum a t r = 1.48 A., represents the [C=O C-0 f C-C] distances. It occurs a t a larger value of Y than the corresponding peak for diketene, but there are twice as many C-C bonds in the metby1 dimer as in diketene. The small peak a t 2.03 A. corresponds to the diagonals in a small four-membered ring. The secondary non-ring distances contribute to the peak at 2.50 8. The larger distances contribute to a peak a t 3.22 A. (similar to diketene and P-propiolactone and unlike the dimethyl dimer) and to peaks a t 3.88 and 4.5 A. All the postulated sonfigurations would have distances near Y = 3.90 A. One of the lactone rings, B has one near 4.50 A. Therefore, according to the radial distribution results from pattern b, the dimer cannot have either a syrnmetric diketone ring or the chain configuration, but can have that of a substituted four-membered lactone ring. Intensity curves were computed for three of the five postulated models; I1 and IV were omitted. No attempt was made to distinguish between these For any one model, and 111, the vinyllactone. variations of the parameters were not investigated, though different configurations due to restricted rotation were. The angles and distances for the various models are given with Fig. 6, with the intensity curves. The critical features of the pattern for structure determination are a shoulder a t s = 4.5,

+

~

1963 OF KETENEDIMER,METHYLKETENE DIMERAND P-PROPIOLACTONE April 5, 1955 DIFFRACTION the relative heights of the two peaks between s = The maxima a t s = 16 and 20 would probably be 8 and 10.5, the width of the maximum a t s = 13.5, removed or decreased if a model with the ketone oxygen unsymmetrically located were considered and of the minima a t s = 16 and 20. Both patterns were compared with calculated in- (compare curves T and K, Fig. 4). The contrast of tensity curves. Type a differs from all the com- the microphotometer data and the scale factor for puted curves, as was anticipated from the radial B are given in Table V. A close check is not exdistribution analysis. Type b is similar in appear- pected since no attempt was made to refine parameance to the diffraction pattern observed for dike- ters in this large and unsymmetric molecule. The acceptable models, determined while drawing tene and @-propiolactoneand could be correlated the microphotometer trace background, agree with with the computed curves.lR The cyclic diketone, curve C, is not acceptable. the results of the visual correlation procedure. AcThe sector curve is helpful here since estimation of cording to the three methods of analysis of pattern the relative peak heights in the doublet is difficult on b the liquid form of methylketene dimer has a subnon-sector plates. All of the chain models, curves stituted four-membered lactone ring configuration, D through H, can be rejected. The variations in rather than that of a 1,s-cyclic diketone or a chain. intensity curves, considering different positions of The parameters are probably similar to those in dithe groups in the molecule, assuming hindered rota- ketene. tion or free rotation, are small compared to the disDiscussion crepancies between them and the observed intensThe visual appearance of the diffraction patterns ity curve. The vinyllactone curves, A and B, agree and the microphotometer tracings agree most adequately with the observed diffraction pattern. closely for dimethylketene dimer, a symmetric molecule. For the other three molecules, all unTABLEV symmetric, these data differed where the visual METHYLKETENE DIMER:INTENSITY CURVEDATA estimation of the pattern is known to be less reliaso so/so Visual Mpr. B/vis. B/mpr Max. Min, ble; the photometer data were found to be consistent with the over-all best models. The micropho(2.10) 2.38 0 tometer results allowed a quantitative comparison 2.65 2.78 1 of intensities as well as SO values, even on unsymme3.90 0.873" 3.94 0.881" 1 tric peaks, to be made. 4.86 2 The confirmation by structural measurements 5.34 2 that the liquid form of methylketene dimer has the 6.20 0.958 0.942 6.10 3 ketene dimer configuration rather than that of the 7.48 .982 7.33 .962 3 dimethyl dimer is interesting since all three dimers 8.40 .985 8.26 ,967 4 can be prepared by the direct dimerization of the 9.05 .958 9.22 .963 4 appropriate monomeric ketenes. No other forms 10.18 10.00 .982 1.ooo 5 of the ketene or dimethyl dimers have been pre11.29 11.35 1.003 0.998 5 pared to date. The mono-enol of the symmetric 11.90 6 diketone form of the methyl dimer is known. 12.83 6 Steric effects'O apparently determine which config13.50 7 uration of the disubstituted dimer can result from 7 15.55 1.017 the direct dimerization. In addition, the possibi17.64 8 lity of an enol form is excluded. Neither of these 19.98 8 0.962 22.42 restrictions applies to the smaller molecules. The 9 question of their configurations involves not only 0,991 23.76 9 0.962 steric effects but energetic considerations among 25.66 10 1.003 the various strained ring possibilities. Evidently 26.74 10 the dienol ring is too highly strained, despite the 1.03" 28.72 11 possibility for a conjugated closed ring system. It 0.98 0.97 Av. is not immediately apparent why diketene should 0.015 0.02 Av. dev. occur only as a lactone, and methylketene dimer Relative Dev. from Specific both as a lactone and the mono-enol of the diketone, height, unit background, contrast, visual mpr. mpr./B or why neither occurs as a diketone. 30 1 -0.180 0.78 The structural results reported by Lipscomb and ,140 35 3 .73 Katz for crystalline diketene are given in Table VI 20 3 - ,032 .50 with the electron diffraction results for diketene and 10 4 ( - .004) P-propiolactone. In the crystal, the molecule is 4 12 - ,008 .20 planar and has the vinyllactone configuration, 111. ,016 21 .24 5 The angles reported by the two experiments are in 17 5 - ,016 .36 excellent agreement. Within the uncertainties, Av . .47 the distances also agree: the over-all size of the Av. dev. .20 molecule is the same within 1.5%. I t is informative to compare the parameters a Omitted in computation of average. found for the lactones with those found for similar (18) Both patterns were compared with the two obtained by Liplinear molecules (Table VI). Divinyl etherI9 conscomb and Schomaker for diketene.17 Pattern b is similar to the one of theirs that is like the diketene pattern. the same s range as their two differ,

Pattern a differs from b, in

(19) L. L. Barricelli and 0. Bastiansen, Acta Chem. Scand , 3 , 301 (1949).

1964

JUDITH

BREGMAN AND S. 13. B.IVER

VOl. 77

TABLE VI COMPARISON OF STRUCTURAL PARAMETERS X-Ray Diketene.

c=c c=o Methoxy C-0 Carboxy C-0 c-c of c-c=c c-c of c-c=o L L L L L L L L

c=c-0 c=c-c 0-cc-c c-0-c 0-co-c c-c-c 0-c=o c-c=o

8.

1.35 1.24 1.39 1.40 1.48 1.46 130" 136' 94 89 ' 94'/*0 €Go

Limits of error Configuration

121" 145' f 0 . 0 6 A. f2O trans Planar

Diketene

1.31 1.19 1.41 1.41 1.52 1.53 130°" 136"" 95 O 89 95 O 81 O 121'" 115" =kO 04 A. 1 4 O

trans

Assumed planar

P-Propiolactone

Electron diffraction, 8. Methyl acetate

Divinyl ether

1.35 1.19 1.45 1.45 1.53 1.53

1.22 1.46 1.36

1.42

1.52 123"

94 O 89 O 94 O

113' 116'

107"

Bo 123O 143O AO. 03 8. f3

124" (1200) 0.03-0.04 A. 3-4 trans cis, non-planar, av. angle Kon-planar, both vinyl Assumed planar C-0-C plane t o O= groups rotated 20' . C-0 plane 25", limitaround C-0 in same diing values 0 and 35" rection

Assumed value from X-ray results.

tains a doubly unsaturated system, as does diketene. It is not planar, and the distances are those found in simpler molecules. The esters, methyl formate and methyl acetate*O are also non-planar. I n addition they both have cis configurations, while the planar lactone may be considered an extreme trans configuration. The two C-0 in each ester are unequal. The angles in the esters a t the carboxy carbons are somewhat different from the 120" expected for strictly trigonal hybridization, consistent with the shortened carboxy C-0 but opposite to the effect found in the p-lactones. Apparently the strain of the planar four-membered ring configuration in the lactones predominates over the tendency in the unstrained molecule for conjugation in CNo

O \ I n view of the structural differences between linear esters and lactones, it is interesting to compare the mechanisms of similar reactions (recognizing, of course, that reacting molecules in solution differ from molecules in the gas phase). The kinetics of hydrolysis of aliphatic esters, @-, and y-lactones have been investigated in detail.21 y-Lactones, are less strained than /?-lactones, and react similarly to aliphatic esters. The cleavage occurs a t the short carboxy C-0 both in acid- and basecatalyzed hydrolysis. In acid solution, the process was found to be bimolecular, the proton reacting directly with the molecule. The mechanism for the @-lactonesis different. In the acid- and base-catalyzed hydrolysis, the cleavage also occurs a t the (now long) carboxy C-0. However, for acid catalysis, the process is the unimolecular opening of

the strained ring; the addition of the reagent occurs later. The deviations from "normal" bond angles and distances shown by diketene and related molecules are part of the more general problem of the structures of small strained ring molecules. The parameters of three- and four-membered hydrocarbon and oxygen-containing rings whose structures are known are listed in Table VII, with the unique features of the structures indicated in the last column. Qualitative attempts to explain them may delineate the problems to be solved. I n the three-membered rings, the abnormalities in bond distances seem to be uniformly toward shorter internuclear distances; in four-membered rings the effect is generally toward longer distances. The shortening in the former has been attributed to the presence of "bent bonds"22in which the line of maximum bonding electron density is outside the direct line joining the nuclei. The lengthening in fourmembered hydrocarbon rings has been attributed to repulsions of the non-bonded ring However, the lengthening of the ring distances cannot be attributed to hydrogen repulsions since such an effect would, in trimethylene oxide, probably result in increased C-C distances instead of the observed increased C-0 distances. The evidence concerning planarity of four-membered rings is not clearcut. Persubstituted rings are reported to be puckered, tetrasubstituted ones planar, and C4He is either non-planar or has a large amplitude out-of-plane vibration. The oxygen-containing rings are planar a t least in the mean. When the two sides of the ring contiguous with a side chain are not identical, the chain lies about 10" off the bisector of the corner. The ring corner angle in four-carbon rings, contiguous with an unsaturated side-chain, is about

(20) J. M . O'Gorman, W. Shand, J r . , and V. Schomaker, THIS (1950). (21) P. D . Bartlett and G. Small, Jr., i b i d , 7 2 , 4867 (1950); S. C. (22) C. A. Coulson and W. E. Moffitt, Phil. Mag., 40, 7th series, 1 D a t t a , J. N. E. Day and C. K. Ingold, J . Chcm. Soc., 939 (1939); (1949). F. A. Long and F. Dunkle, private communication; F. A. Long a n d (23) J. D. Dunitz and V. Schomaker, J . Chcm. Phys., 2U, 1703 L. Friedman, THISJOURNAL, 73, 3602 (1950); F. A. Long a n d M. (1952). Purchase, i b i d . , 7 2 , 3267 (1950). J O U R N A L , 73,4222

1965 April 5, 1955 DIFFRACTION OF KETENEDIMER,METHYLKETENE DIMERAKD 0-PROPIOLACTOIUE

PARAMETERS OF

c=c

Compound

C yclopropaiie Spiropentane

2

Cyclopropene Ethylene oxide

TABLE VI1 THREEAND FOUR-MEMBERED RINGSz3

Distances, k.

c-c

I . 525 1.515 f 0.02 1 . 5 1 f .04 Ci-Ct 1 . 4 8 f .03 Ci-Ca 1.286 f 0 . 0 4 1.525 f .02 1.472

1.568 f .02

Tetraphenylcyclobutane

1.555 f .02 1.585 f .02 1.59 1.60 1.55 =k 0 . 0 2 1.56 f .03 1.54 f .03

Per-substituted cyclobutanes Methylenecyclobutane 1.34 i 0 , 0 2

Methylcyclobutarie Dimethyl ketene dimer Trimethylene oxide P-Propiolactone

1.34 f 0.03

-

1.56 1.54 1.56 1.54 1.54

f A f

f f

Unique features

Angle4

LH-C-H = 118 f 2" LC-C-H = 116 f 2" LC2-C3-CI = 8 1 . 5 2" LH-C-H = 120 i 8" LC=C--M = 152 f 12' LH-C-H 116'41' LC-0-C 61' 24' LH2-C-C = 159" 25' Non-planar ring on average LH-C-H = 114 f 8" Ring planar

+

1,436

Cyclobutane

1-Methylcyclobutene

c-0

Xon-planar rings LC-CCH2-C = 9 2 . 5 f 4'

LC=C-C = 93.5 f 3" LCHa-C-C = 125 f 4' .03 ring Ring may be non-planar .06 chain LC-C-CH3 = 118" .05 ring LC-CO-C = 93.5 f 6' .05 chain LCH3-C-CH3 = 111 f 6' .03 1.46 4 0.03 LC-0-C = 94.5 f 3' LC-C-0 = 88.5 f 3' See Table VI

C-C short C-C short C=C short C-C short

C-C long C-C long C-C long C-C long L s a t C=C L s a t C=C C-C long C-C long L s at C=O C - 0 long

c-0 long L s a t C=O

Diketene

93", whereas for a planar unsubstituted ring i t is 90". Accumulated evidence points to considerable delocalization of the electrons in the planar strained rings. With the bonding orbitals becoming more alike, greater dimension symmetry in the ring results. I n both ethylene oxide and trimethylene oxide the ratio of distance C-O/C-C is closer to unity than it is for linear molecules, although in the former the C-C is short, and in the latter the C-0 is long. I n addition, unsaturated side-chains seem to interact with the four-membered rings appreciably. The lengthening of C-C in saturated four-carbon rings is least in the cases of dimethylketene dimer and methylene cyclobutane. One may argue that in f i e three member rings, the necessity for "bent bonds" leads to a short C-C, with the C-0 remaining nearly normal; were one to measure along the bonding orbital axes, the CO would also be longer than 1.42 A. On the other hand, in trimethylene oxide, delocalization permits the ef-

See Table VI

L s at C=O

fects of non-bonding repulsion to be absorbed in lengthening the C-C. Acknowledgments.-We wish to thank Professor J. R. Johnson of Cornell University and Dr. R. Sliegel for the samples of dimethylketene dimer and methylketene dimer, Professor A. T. Blomquist of Cornell University for the sample of ppropiolactone, and Drs. L. Diuguid, F. Fedoric and 11.1. Passer for distilling some of the samples. The progressive digiting method for computing sums of products was worked out by J. B., with the help of Dr. L. H. Thomas and Mr. Eric Hankum whose assistance is most gratefully acknowledged. The kindness of the Watson Scientific Computing Laboratory in making computing machines available is also gratefully acknowledged. J. Bregman wishes to thank Cornell University for a Sage Fellowship which was awarded during the course of this work. ITHACA, N. Y.