produced when the yellow complc.; is irreversibly transformed into the lawnclcr coniplex. If thc appearance of additional bands in this region is due to cyanide bridging, then i t ~ w u l r lfollow that an increase in extent o f bridging should be mirrored by changes in relative intensities. lvhether or not such is the casc here remains to be substan tinted,
Xdditimal studies which might reveal more about the structural featurcs of the lavender complex wcre unfortunately precluded by failure to find a solven t which would dissolvc sufficieiit aniounts to enable measureinent of inolecular weight, tlipolc iiioiiicii L or conductivity .I\\
\RIIOR, l f I C l I l G \ \
[ C C l X r R I l 3 K ~ r I O XFROM TIII: L)CPARTlIEST OF C I I C M I S l ' R Y , hfASSAClIKSCTTS I N S T I T U T E O F
TECI~SOLOCV]
Structure of the Reaction Product of Phenylasetylene with Iron Pentacarbonyl BY J. R. LETO,\ED F.A , COTTON I < ~ c s i r ~SEPTEMBER r, 8, 195,q :\ coinpiiuiitl first prepared by Jiines, IVailes :inil IT'hiting in 1 11:~s1)c.cn filrtlier stuilietl. I t i q conclutletl from analytical rlat:t t h i t its foriiiuki is CgoH1?04Fe and not C y o H ~ ~ O ~ P c. study (if the infrared spectrum and the fact t h a t it has :I Detu dipole moment of -3 Dcbyc units 1e:id t o t h e propriwl of a structure it1 wllicli :i riiplien~lcyclo~,entndienone is bonded by 7: electroiiq to :in Fe(C0)3 group.
Introduction I n 1933 Jones, U-ailes andL\-hititig' reported the isolation of a coinpound with the formula CroHlaor Is04Fe(I), their analytical accuracy being insufficient to distinguish between 10 or 12 hydrogen atoms. This coinpound was obtained in very small quantities by adding Ni(CO)4 to a inixture of Fe(CO):,and phenylacetylene in ayucous ethanol a n d acetic acid. JYithout the Ni(CO)* no reactioii occurred, but upon addition of Ni(CO), the usual carbonylation reaction producing CaHsC(COOC!Hs)-CH2 proceeded essentially quantitatively antl traces of I wcre found :xinorig the products. Jones, ct ai.,proposed that I is actually C2aHloOdFe with structure a , Fig. 1. Since, however, transition metal acetylides are coillinonly unstable, whereas I is e s t r e i d y stable, we have considered that I might have the structure b (Fig. 1)' and thus be CroHluOIFe. \Yith this in ,ilincl, we felt that the coinpound merited further study. L1-e have confinned its esistence, preparing it by the same irrational procedure used by Jones, P / nl., with slight inodifications and h a w tleteniiiiietl a nuiiiber of its properties which bear on its coiistitution :mtl structure. I t appcars to us t1i:it thc ciklciice now available is inconsistent with structure a , but consistent with structure 11. Experimental Preparation.--This IKLS cnrrictl o u t in c~seiitinllj~ tlie m:inner described b y Jones, IYailcs antl IYliititig.l Thr. yield is increased simiewh:it (frriin 3.:2 to 6.0(,:,) by employing lotiger reaction tiincs, ant1 especially if the material is : t l I ~ i ~ v e to d crystallize overnight frijiri the cooled solution. 111 :I typicnl r u n 2.2 g. o f I is obt:iined from a reaction inisture consisting o f 19.4 g. o f Fe(CO)j, 4.9 g . of Ni(C0); ~ i n d25 g. of phenylncetylene, wliich had been held a t 70 fiir 3 hr. Altogether a stock of 6.5 g. of I mas accumulated from 4 runs. The priitluct was purified as described by Jones, ff (d.,and had :I melting point o f 221-222' (222" reportedL). Purification by Chroxnatography.-,A colunin o f 200 g. o f :iluiiiina was thoroughly w:ished with benzene and a solu-
tion of 0.2 g. of I in benzene WAS put on the column. Blution was carried out sloivly by using a benzene-chloroforni mixture, gradually increasiug the proportion of chloroforin. T o separation of the hand occurred. Elution was continued using benzene-acetone mixtures a n d again no separatioii was observable. Elution was then completed with pure :icetone from which I was crystallized and found to h:ive the same melting point as before. Magnetic Susceptibility.-The bulk susceptibility IWS measured on the solid material using a Gouy balance. Material which as not rigorously purified shortly before inensuremerit antl carefully protected from light was fouiitl to be far less diamagnetic than would be expected and o i i some occasions slightly paramagnetic. T h e i n o h r stisceptihility obtained on the most highly purified :inti carefully handled sample was -40.0 X 1 0 F c.g.s.p which is still less than would be expected from Pascal's ccinst:ints. Nuclear Magnetic Resonance.-The proton reson:ince pectrum o f a saturated solution of the iron coinpiiunti i n icetone-& was taken in an effort to decide the nurnbcr ant1 kinds of protons in the molecule. The liighest power level of a T-arian 4300B high resolution spectrometer was usetl, : i n d the spectrum wa? scanned a t various speeds; only one weak resonznee was ohwr\-ctl :it + I .(is p.p.ni. froin water :ind is without m y tloul)t due t o the phenyl protons. 111 our experience coinpountis cimt:iining acetylenic or olcfinic protons have :I resonance in the region $0.7 to + I .(l I).p.in. from w a t e r , but a careful search i i i this region (and all regions) yielded notliiiig. The phenyl resonance peak height (or area) in the s:ituratetl solutiiin of the iron conipound and the peak height (or area) o f the cyclopentadienyl Iirotons nf ferrocene (ridded in mole for rnole amount t o t h e solution) n-ere the same ivithin csperirnental error-that is 1 0 l~rot(itisper riiole in ferrocene gave the same area under resonance peak as all the observable protms per inc~le (pheiiyl) in the iron cornpound. Xlthougli this indicat cs that t h e 3 other prntons :Ire not "irlaskcd" b y t h e plicnyl rcsonaiice, it ~ I ~ niitJpoqitively C S disprove tlieir esistence. T h c ol,scr\.etl peak is sulficicntl~~ weak th:it in our judgmeiit :mother peak ' / R it intensity could escape detection. Infrared Spectra.-Infrared spectrx \vere taken 011 saturated solutions in CS2, CIIClr and CCls anti OII t h e solid mullet1 in Nujol and licxachlr~ri~hutndieneand pressed in KI3r. r s i n g 0 . i mni. cells, the solutirins wcre all too dilute to rere:il a n y but t h e strongest Ixincls. The spectrum o f pure liquid phenylacetyleiie \vas also run for coinlxirison. 1111 of t h e IJbserVetl bands are listctl in Table 111. Perkin-Elmer model 21 double beain spcctroiiieter cquipped with a rocksalt prism \ws used, employing both scanning c:inls to obtain the best poisible resolution :iround 2000 cm .-I. e Moment.-The dipole iiionieiit icnzene solution was ealculated in ureineiits o f the total pohrimtion m t i o i i ( i f thr. inoleculc. The t l i -
June 20, 1939 STRUCTURE OF REACTION PRODUCT OF PHENYLACETYLENE WITH IRON PENTACARBONYL 2971 TABLE I EXPERIMEXTAL
Solution
S I
DATAAND RESULTS O F DIPOLEMOMENT hfEASUREMENT'
s 2
C,?
d
TP
?1D
P
EP
Benzene I .oooo o .oooo 2.2741 0.5734 . .. ... I . 4975 ... .. . ..,,... I (satd. in benz.) 9.99818 ,00182 2.3011 0.5742 326 f 17 1.4982 130 =k 13 3.11 0 . 2 I (satd. in CI-ICIS) 0.99432 .00568 .... 1,4678 ...... 1.4453 11G & 2 ....... Chloroform 1 ,0000 o m .., , 1.4717 . ... , , 1.4400 ..., .. ...,.,. = dielectric constant of solutions ( t o =k0.0003); d = dena ATl = niole fraction solute, NQ = mule fraction compound; ,iity cif sulutions ( t u 0.0002 g./cc.); T P = calculated total molar polarization of solute; nD = refractive index of solution; gP = electronic polarization of solute calculated from n"; IJ. = dipole moment (Dcbycs) = 0.0128 ~ ( T PaP)T.
-
electric constants of pure benzene and a saturated solution of I in benzene were measured on a high-frequency capacitance bridge designed in the Laboratory for Insulation Research a t M . I. T., employing a self-filling, thermostated cell with cylindrical electrodes. Densities were determined by means of a calibrated 10-cc. pycnometer and a Mettler balance accurate t o 0.02 mg. The densities given a r e in error by no more than 0.0002 g./cc. Indices of refraction were measured on an Abbe Refractometer, using t h e sodium-D lines. All measurements were carefully therniostated to 26.0 i 0.5". T h e mole fraction of the solution was determined by weighing the residue after evaporation o f t h e saturated solution. D a t a are collected in Table I. hlso included are data checking the electronic polarization in chloroform which is a better solvent for the compound. An attempted series of measurements in dioxane was not successful because of slow decomposition of I in even carefully purified dioxane.
Discussion The principal lines of evidence bearing on the composition and structure of I are: (1) analytical data, (2) its dipole moment and (3) its infrared spectrum. Before proceeding to discuss these in detail a few other observations may be noted. \Ye have confirmed in general all of the chemical and physical properties of the compound which were reported by Jones, LVailes and Whiting. The chromatographic homogeneity study confirms the fact that I is a single compound and that it is pure as obtained by the procedure of Jones, et al. The compound is doubtless diamagnetic. It might be expected to have a negative susceptibility of the order of several hundred X c.g.s.u. whereas we obtain only -40 X lop6 c.g.s.u. on the purest samples. This might be due to the presence of a low-lying paramagnetic state of the molecule, but it is more likely due to photochemical or possibly oxidative decomposition producing traces of paramagnetic substances such as iron oxides. Jones, et al., observed the photochemical instability of I, and we have noted that on standing in strong light a greenish cast develops and that such visibly greenish specimens are appreciably paramagnetic. The infrared spectrum of phenylacetylene shows a very strong band a t 3330 cm.-' due to the =C-H stretching mode. Jones, et a/., reported that there was no such band in their infrared spectrum of I. We have confirmed this on spectra under a variety of conditions and thus are in agreement that a structure having two C6H;C=C-H groups coordinated via the C=C bonds to an Fe(CO)4 moiety is not admissible. Such a structure would also require the iron atom to have an effective atomic number of 20. The n.m.r. spectrum does not provide any firm evidence for or against either structure in Fig. 1. The peak due to the phenyl protons is weak even in (CD,)?CO the best solvent we have found, and it is impossible to say
whether another peak one-fifth its intensity is truly absent or simply undetected. 1. Analytical Data.-If we consider the the0 retical analyses to be expected for C:oHlo04Fe and C20H1204Fe, Table I, we see that dlfferences in the carbon, oxygen and iron percentages are too small to be of any use in reaching a decision between the two structures. However, the hydrogen percentages differ appreciably. Table I records the results of numerous analyses performed indepen dently by several different laboratories. It is clear that these results argue strongly for the formula C?oHI2O4Feand hence again st structure a .
CP I
C
11 C
P
n
C
0
(a)
(b)
1
Fig. 1.-Possible structures for C ~ O H I O ,?04Fe: ( a ) the structure proposed by Jones, Wailes and LVhiting assuming the formula CsoHloOJ?e; ( b ) the structure proposed in this work for the formula C*OH1?O4Fe R', R2, R 3 and R 4 represent two hydrogen atoms and two phenyl groups the distrihution of which we cannot specify.
2. Dipole Moment.-The dipole moment of the compound was found to be 3.1 =t0.2 Debye units. The significance of this datum is clear without much comment. The relatively low accuracy is unavoidable because of the low solubility. The concentration of the saturated solution in benzene was so low that no attempt was made to measure more dilute solutions for the purpose of extrapolating to zero concentration as is usually done. Since i t is generally found that apparent moments increase with decreasing concentration of the solutions measured, the result obtained can be regarded as a lower limit. The moment found may be compared with the moment of 3.3 Debye units found for (7r-CjH;)Mn(C0)3 by W e i d which suggests that it is reasonable if the compound has structure b. I t appears quite impossible to reconcile it with structure a since this is centrosymmetric and uncorrected-for atomic polarization could scarcely (4) E Weis?. E u n o i g olidrin C h r m
287, 22'1 ( 1 0 5 7 )
2972
J. K.LETOAND F. A . COTTON
1701.
s1
TABLE I1 TS FOR THE C O M P O U N D , Aualyses, 'y0-
Source
RIUI.
1. Calcd. 2. Calcd. 3. Jones, Wailes, LYhiting
W . "
370, 1 8 372 I5
...-
S. hl. Xagy, M. I. T . 5 . Schwarzkopf Labs., N. Y . 6. Present authors 7. W. Manser 4.
.,..
c
Ice
04.90 (i4.d-l M ,4 61.5 61.2 G I .2'
1 5 , 09 15,OI
~ . -
-
II
.
2,723 :3,i250 3 0 :1. I I'. .:I4 :%, 33
1.7.5
...
C2JIlo04€~e C~~HI~O~I'C Cl?i.4TT,".,03.sF'eb
C17.11111.d33.7Fed C1~ . I H 1 2 . 1 0 ~ . o F e ' .... 60.80 ... 3.63 C1~H13.30d.iFe~ The compound is not sufficiently soluble in camphor t o make possible a determination by the Rast method and it9 low solubility in benzene and other solvents discouraged attempts t o determine an approximate molecular weight cryoscopically or ebullioscopically. We are t h u s assuming t h a t the molecule is not a dimer or other polymeric form of the basic unit which contains only one iron atom. The high iron and low carbon analyses suggest a n incomOxygen obtained by difference, plete decomposition of the sample, leaving some type of iron carbide in the residue, from which the iron percentage was calculated. I n one analysis performed by M r . D. L. Guernsey of t h e Metallurgy Dept. of M. I. T., a sample was decomposed at 1250", giving a percentage of 62.5 zk 0.5%. Oxygen 1 7 . 5 9 7 b~y direct analysis; total 98.S'%. a Iron by direct titration. Result of five determinations, four by titration of the iron(II1) thiocyanate complex with 0.1 N disodium ethylenediaminetetraacetate, one sample b y a Zimmerman-Reinhardt permanganate titration. Care must be taken to decompose the samples colnpletely in concd. HCI-HNO3, and then t o adjust the pH to a proper range for titration. f Calculated using C = 62.5, F e = 15.2, 0 = 17.6, H = 3.3. Total 98.6y0, 0 Direct oxygen determination gave 18.05 yo; forr~iiilacalculated using C, 64.5; Fe, 15.2. .,..
16.7" 13. 2e
Cumpiited empirical formula
..,
TABLE 111 ISPRARED SPECTRA O F KRr
Phenylacetylene
3350(vs) 3 120sh(m) 3105(m) 3075sh (in) 2104(m) 1970B(w) 189OB(w)
M O B ( w) 17 60 B (w ) 1675B( w) 1604(m) 1581(in) 1496(s) 1451(s)
1395(w) 1335(w) 1285( w) 1250B (m) 1180(w) 1160(w) 1100(w) 1073(m s ) 1028(ms) 9 18(111s) 883(w) 840(W)
75S(vs) 690(vs) 655(v5) R , brnad, lon -frequency
P H E N Y L A C E T Y L E S E A T D ITS . I D D U C T \VITH I R O N C A R R O S Y I . "
-3155(~) 3090(\r)
h
1885(vm) 20 90 ( v s) 2026( Vi) 2008(Vi)
2085(vs) 2024(vs) 2002( vs)
2083(vs) 2022( vs) 2000( v,)
1800(vw) 1740(vw) 1700(W) 1F20(Vi)
1700(w) tFl8(vi)
1700(w) 1615(vs)
1506(w)
1506(w) 1453(m) 1443(m)
1507(W) 1457(m) 1445(m)
1333v w )
134O(w)
1213(m) 1188(w) 1156(vw) 1103(vw) 1O88(w) 1030(R )
1212(w) 1187(vw) 1156(vm) 1103( vw) 1084(w) 1030(w)
886(w) 852(w) 772(m) 753(s) 749(s) G97(5) 693(s) 652(in) 648imI s. strong, 111, Inedlum; w, weak; v, very; wlc
884(w) 855(w) 776(s)
be so large as to account for a11 of the measured 7P - FP. 3. Infrared Spectrum.-Various features of the infrared spectrum appear to be inconsistent with
121O( \I )
1103(v w ) 1082(w) 1027(w)
7701'V) 747(m) 692(m) ms, medium strong
I,
ShoLTi pronounced broadening on the
structure a but consistent with structure b. Let us consider first the region between 1900 and 2100 cm.-I. I n all cases (with the possible exception of the C1ICl3 solution) three strong bauds are ob-
June 20, 1959 STRUCTURE OF REACTION PRODUCT OF PHBNYLACETYLENE WITH IRON PENTACARBONYL 2973 served. This is perfectly consistent with structure b since the symmetry of the molecule cannot be higher than Cs and the Fe(C0)3 grouping would therefore produce three infrared-active CO stretching modes. In structure a it seems unlikely that the D d h symmetry of the Fe(C0)h grouping would be significantly disturbed by the remote phenyl groups. If this assumption be correct, the Fe(CO)4 grouping would produce only one infraredactive CO stretching mode. The linear C6H6C= C-Fe-C=CC& would produce only one infrared-active C r C stretching mode which could occur as low as -2000 ~ m . - ’ . ~Thus one would expect structure a to produce only two bands in this region contrary to observation. Another critical feature of the spectrum of I is the strong band appearing in the range lGlS-lG47 cin.-l depending on sample state. Phenylacetylene has two bands of medium intensity near this region, a t 1551 and 1G04 cm.-l which can probably be assigned as phenyl ring deformation modes. I t does not seem reasonable that changes occurring a t the carbon atom second removed from the ring should drastically alter the number, position and intensity of such bands and this is supported by the fact that the phenyl hydrogen wagging bands between 600 and SO0 cm.-’ are virtually identical in phenylacetylene and the complex. This reasoning thus makes structure a implausible. However, the appearance of a very strong band in the ketone CO stretching region is only to be expected for a compound with structure b. The structure b and the bonding which we are proposing for I are not without some precedent. The most direct analog is the duroquinone-Fe(CO), compound of Sternberg. Markby and Wender.6 Also, the (1,3-b~tadiene)-Fe(CO)~ compound of Reihlen, et ~ l . and , ~ the similar (1,3cyc1ohexadiene)-Fe(C0)s compound prepared by Hallam and Pauson* and the structures proposed for these by Hallam and Pauson are quite analogous to structure b for CSOH1204Fe. Finally, we learned ( 3 ) T h e C E C stretches in silver acetylides occur at -2000 cm. -1; L. Luttinger. private communication. (15)H. W. Sternberg, R. M a r k b y and I. Wentler, THISJOURNAL, 80, 1009 (1958). (7) H. Reihlen, A. Gruhl, G. Hessling and 0. Pfrengle, A n n . , 482, l e 1 (1930). (8) B. I?. Hallam and I’ I,. Pauson, J . ~ ‘ I w i u S u r . , 043 (1038).
recently9 that Sternberg, Markby and Wender have isolated a compound in which a a-CsHbCo grouping is bonded to tetrasubstituted cyclopentadienone, for which they independently assumed a geometry similar to what we are proposing for C?oHlz04Fe. If I is actually CzOH120dFe with structure b as seems reasonable from the preceding discussion, then the failure of Jones, Wailes and Whiting to obtain it by several “rational” syntheses is easily understandable. Their “rational ”syntheses were both intended to produce structure a and started with either CsH5C=ChlgBr or CeH&kC-I so that a compound containing twelve hydrogen atoms could not normally be expected to result from their reactions with Fe(C0)j or NazFe(CO)d, respectively. Confirmation of the structure proposed here might be obtained by syntheses of the compound by reaction of one of the iron carbonyls with the appropriate diphenylcyclopentadienone (whichever one that may be) except that none of the three isomeric forms has been isolated and unsubstituted or partially substituted cyclopentadienones are generally thought to be unstable. We do not consider it unlikely, however, that a diphenylcyclopentadienone system could be stabilized by combination with the Fe(C0)3 group. For example, Wilkinson10 has recently shown that cyclopentadiene (C5H6) can codrdinate to metals via its two double bonds, whereas free C ~ H itself G is unstable in the sense that it readily dimerizes. Perhaps the best possibility for rational synthesis of a compound similar to CzoHlzO$e would be by the reaction of the well-known tetraphenylcyclopentadienone (tetracyclone) with one of the iron carbonyls, although the stability of tetracyclone against polymerization may also mean that it will be unable to form the required pair of bonds. The matter is under study in these Laboratories. Acknowledgments.--We thank Dr. R. W. Fessenden for assistance with the proton resonance measurements and Mr. William Westphal for the use of his dipole moment apparatus. This work was supported by the United States Atomic Energy Commission under Contract AT(30-1)-1%5. CAMBRIDGE, MASSACHUSETTS (9) H. W. Sternberg and I. Wender. private communication, in course of publication. (10) G . Wilkinson. Presented a t 133rtl N a t i m n l Meeting of T h e American Chemical Suriety, San I’rancisco, April, 1 % %
