S. H. BAUER
3 104 [CONTRIBUTION FROM THE
D?PARTMENT OF
Vol. 6 9 .
CHEMISTRY, CORNELL UNIVERSITY ]
An Electron Diffraction Investigation of the Structure of Difluorodiazine BY S. H. BAUER Introduction When one considers the compounds which could be formed between nitrogen and fluorine, our limited experience suggests the following four : NF3, N2Fz,NzF4 and N3F. Compounds of higher molecular weight would require nitrogen chains involving more than three such atoms, and to our knowledge, attempts to prepare longer chains of nitrogen atoms have not been successful. Fortunately, small samples of fluorine azide and difluorodiazine were available to us for structure determination ; tetrafluorohydrazine and nitrogen triflu.oride were not. However, the structure of the latter has been established. The infrared absorption spectrum of nitrogen trifluoride was first obtained by Bailey, et aL,l who made a provisional assignment of the observed fundamental frequencies. The molecular symmetry is CW; and the N-E? separation, as deduced from the force constant of 0.410 megadynes/cm. by means of Badger’s rule, is 1.45 B.; the L F N F is 110’ or somewhat hrger. Dr. V. Schomaker communicated to us his elecJron diffraction results on NF3: N-F 1.37 *: 0.02 A.; L F N F 102.5 * 1.5’. We thank him sincerely for these data.
“At ordinary temperatures difluorodiazine is a colorless stable gas with an odor like NOz. It condenses to a colorless solid a t about - 110’ (v.P. 100 mm.).” Only qualitative analyses were made on either compound; both were shown to consist of nitrogen and fluorine only. Upon removal of the excess nitrogen, molecular weights yere determined. For N3F (61)) 61.6 and 62.5 were found, while for N2F2 (66))66.2 and 6S.0 were obtained. The high values in both instances were accounted for by assuming slight contaminations-in the case of N3F with NzFz, since i t was practically impossible to avoid some decomposition in spite of the rapidity of the measurements, and in the case of NZFZ with NFI, probably introduced through the action of excess fluorine on the fluorine azide; thus N3F
+ FP= NFa + No
In interpreting the electron diffraction data, these complications had to be taken into c ~ n s i d e r a t i o n . ~ The author is greatly indebted to Dr. Haller for the samples of NSF and NZFz, and for aid in handling the compounds.
The Electron Diffraction Photographs Fluorine Azide.-Diffraction photographs Remarks Regarding the Materials Available were taken of a freshly prepared sample In contrast with nitrogen trifluoride, which has a t about - i s o , a t a pressure of 150 mni.).(kept Albeen known for more than eighteen years, fluorine thoygh the contents of bulb were distinctly greenazide and difluorodiazine have been prepared only ish-yellow a t the beginning of the run, twenty recently by Haller working under the direction of minutes later the color had faded considerably. Professor A . W. Browne3; the former by treating The resulting patterns were all alike, and were in a stream of nitrogen, hydrazoic acid with fluo- qualitatively indistinguishable from those obrine, for which they suggest as the most probable tained when electrons were diffracted by NzFz, reaction whereas a comparison of the theoretical curves for 4HNz + 2Fz = NHiF Nz 3N3F the most probable structures indicated that a large and the latter compound through the decompo- difference should be observed. We therefore consition of the fluorine azide, according to the re- cluded that the brass tube leading from the sample flask to the nozzle probably catalyzed the conaction version of the unstable fluorine azide to the more 2NaF = 2 S c + NzF: stable difluorodiazine.” The following characteristics of the compounds are Difluorodiazine.-Photographs were taken on quoted from HallCr’s dissertation. “Fluorine Eastman Kodak Co. Commercial plates of a azide is a greenish-yellow gas a t ordinary tempera- sample kept at room temperature, under an initial tures; it liquefies a t -82’ when distilled with pressure of 81 mm. The accelerating potential nitrogen, and freezes to a greenish-yellow solid a t was 43.76 KV, and the nozzle-plate distance 19.12 - 154’. When the liquid is vaporized, explosion cm. Five maxima were observed; the general generally occurs. .In the gas phase, a t pressures appearance of the photographs is as sketched in under 200 mm., N3F decomposes, slowly a t room Fig. 1 (V); it is indeed doubtful whether the temperature and very rapidly near the boiling slight shoulder t o the right of the fourth maximum point of water, t o produce nitrogen and NzFz. is real; subsequent study of the photographs con(1) C. R.Bailey, J. B. Hale and J. W. Thompson, J . CAem. P h y s . , vinced us that this peak is somewhat broader than I, 275 (11937). the others, but is essentially symmetrical. No ( 2 ) 0. Ruff, .1. Fischer and F. Luft, Z. UFZOYE. Chcm., 172, 417
+ +
(1928). (3) J. P. Hallpr. Doctoral Dissertation. Cornel1 University. Srptember, 1942.
(4) Assuming NFa t o be the sole impurity in the NnFs. the particular sample for which the measured molecular weight was 68, had a5 much as 20yo of the impurity.
ELECTRON DIFFRACTION INVESTIGATION OF DIFLUORODIAZINE STRUCTURE
Dec., 1947
3105
cation that some impurity, very likely NFs, was present. Analysis of the Photographs The SO = (4n sin 6/2)/X values computed from visually determined diameters are tabulated (Table I) ; in the case of the fourth peak, the edges rather than the maximum of density were measured (the value 14.0 is interpolated). It is likely that the tendency to do this resulted in values which are too low for the fourth minimum and too high for the fifth minimum. X radial distribution curve (Walter and Beach method, Fig. 2, R.D.) 1.42
2
4
6
8
10
12
14
16
18
3.64
2.30
ii
20
S+.
Fig. 1.-Relative
A -
2
intensity curves for N2F2.
Somewhat idealized sketch of the visual appearancr of the photoyaphs. E:: N = N 1 28 A,. S-F 1 45 L S N F 120' (czs trans) E1 : 1 28 1 45 110" (CLT lrnm) J: 1 28 1 45 112O (czs fruils) : 1 28 1 50 1 1 G " (cis traits) L: 1 28 1 55 116" (cis Irans) \':
+
CY
+ + + +-
molecular weight determination was made on the particular sample used. However, since i t was prepared parallel to one for which the molecular weight data were obtained, there was thus an indi-
4
6
S
10
12
14
16
18
20
S+.
Fig. 2.-R.D. radial distribution curve for iVeF2. The others are relative intensity curves for N-F
1 . 4 5 A.
Q:
SF3
X:
S-S +/"
(planar)
1 15
Y:
\F - +/F N=N\F
(planar)
1 15
Q: X: Y:
(SYWZ
Csr)
LFNF112' N=N 1.2413. N=N 1.16
L F S F 12%" LFNF116"
S. H. BAUER
3106
Vol. 69
TABLEI Soalod. /Sobs. so
C
1
4.49
(1.006) (0.925)
2 9 i
3a 4
4a
(4) 4b
5 9
Model H
T
C
(1.129) (0.944)
1.052 1.050
(0.9251 ( ,895) ,949 ,959 ,999 ,992 1.017
1.055
C+T (1.074) (0,901) ,919 ,974 ,997 1.014 1.028
1.002
,988
1.034
1.018
.987
0.992
0.991
0.937 ,979
0.960 .984
.936
,926
.994
.990
966 .959
.Y12 ,959
.972
,965
,0237
,0200
jI1.X
1.068
12.2t (13.64)
1.012
14.0 (14.72) 16.66 17.83
1.051
1.036
1.029
1.021
1.Ul6
U.954
0.956
0.968
,976
:998
0.955 .993 1.004
Average Av. dev. Deduced S=S: S-F: Assumed i S S F :
1.010 0.0298 0.0306 1.28 1.29 1.28 1.46 1.46 1.46 +112O 4
1.004 0.0369
.984 ,986 ,0212 1.26 1.43
S+-+F N=N, /' F*-----*
froni the cis a t B..iO A. is not resolved F from this peak, but appears as a shoulder) ; and ,N=N \ distances 3.64 -4.-corresponding to F+-----+
F
(transj. X large number oi theoretical intensity curves were compuled for the cis and for the trans forms of F--h-=N-F, in which the ratio of (N--N/ N-F'I was varied from 0.823 to 0.985, and the L N N F from 112 to 125' (Fig. 3 ) . The acceptable models are G, H and J ; for the latter the trans modification gives the better fit, whereas for the first two, the cis form is to be preferred. X.Iodel:j E, L and F may be excluded due to lack of qualitatis-e and yuantitative agreement with ob-
c
A
n
F
( 1 830
,949 .99ij
.996
I ,027
.0296 1.27
1.27
1.44
1.44
t- 116'
shows prominent atomic spacings a t 1.42 A,corresponding to N-F distances; 2.30 A.-corre/x , sponding to distances (the contribution
0.900 0.950 1.OUO (N-N)/N-F) +. Fig.3.---The held uf parameters covered in the coinputation of intenxdty curves for cis and trans F-N=X-F.
Model F
,974
,994
1.005
T (1.103) (0 .925) ,926
,977 .970
9.21
,987
Model G
C
,966
7.7fj
07
C+T (1.1091 (0.938)
C+T (1.020) (0.893) ,924 ,953 ,969 ,993 1.011
,977 ,976
4.66 fi
3
Model J
T (1.180) (0.955) ,960 1.009 1.012 1.040 1.073
Max. Min.
(0.914) ( .863)
,926 ,931
.0298 1.23 1.24 1.44 1.16 f- 116"
.958 ,0255
1.24 1.4*->
+
+
T (1.146) (0.944) ,970 1.027 1.023 1.045 1.069 1.075
1.104 1.074 1.110 1.055 I).0390
1.35 1.4.7 1.90"
servation (the first and .fifth maxima in E; the third maximum and shoulder in L; the region of the fourth maximum in F). Since we could find no chemical reason why in the preparation either form should predominate, and since the differences between the computed curves, both qualitative and quantitative, are within the limits of discrimination which we believe are justified, equimolar mixtures of the cis and trans forms are shown in Fig. 1. The qualitative agreement of the various models can best be judged from comparison of these curves, and the quantitative fit from the values in Table I. Insofar as the general form of the pattern is concerned, curve G is most satisfactory; the others do not give as good a fit either with regard to the detailed structure of the third peak or in the region 17 > s > 19. The quantitative agreement between the measured and computed SO values is not as good as one would wish. We are thus led to the conclusion that the impurities present did affect the pattern to some extent. To ascertain the disturbances which would be introduced by nitrogen trifluoride, several theoretical intensity curves were computed for that molecule, in which the symmetry Carand an F-N distance equal to 1.45 A. were assumed (one of these is curve Q in Fig. 2). Superposition of curves Q and G demonstrates the effect such an impurity would have on the quantitative fit. Unfortunately, due to the fact that only small amounts of NzFa were available (1.5 X mole) and that the preparation involved making small quantities of N3F, which exploded three out of every four runs, we could not retake these photographs nor obtain patterns with more rings. The assigned limits of error are consequently large. The most probable values deduced (Table I) are x=s 1 2.5 =k 0 04 A. \
--F
LYSE;
1.44 0.04 \ 115 *
Dec., 1947
ELECTRON DIFFRACTION INVESTIGATION OF DIFLUORODIAZINE STRUCTURE 3107 TABLE I1
Bond
B-I:
C-F
S-F
s-c1 N-Br 0-F
Co6rdination
Trigonal Trigonal Tetrahedral Tetrahedral Tetrahedral Tetrahedral Pyramidal 115' LN=N-F Pyramidal Pyramidal LONCl 116' L OSBr L F O F 105' LFOX 105'
predicted* Dist.
Obs.dist. 1.30 =t, 0.02 1 . 2 9 * .02 1.43 * .03 1.396 1.36 * .02 1.34 * .02
Ref.
10 11 13 14 15
1.35
*
16 This paper
1.3'7
.02 * .04 1 . 7 7 1 .02 1 . 7 4 1 .02 1.95 * .02 2 . b 1 .02 1.37 1.44
1.41 1.42
* *
.05 .05
9
1.3912
8 8 8
8
1.73 1 86
8 8
1.41
smaller than those computed by means of the Schomaker-Stevenson relation for single bonds,8 in the case of the carbon fluorides and nitrogen halides, the observed distances are larger than the computed ones, Table 11. Indeed, the rather 'F is almost satisfactory were one to base his judg- strained postulate proposed for the fluorinated ment on the intensity curves alone (Figs. 2 and 4). methanesl4 involving a primary ionization folThus, models X, T and Y give curves which look S much like the observed pattern except for details 125 in the region of the third and fifth peaks; the quantitative agreement is not as good as for the 1 F--N=N--F models. Model W is definitely not 0 sat isf actory . K v
These check very well with the observed peak positions in the radial distribution curve. It is interesting to note that although the con+ /F is chemically not tenable,6 i t figuration g = N
