July 20, 195.5 / C O S T R I B U T I O S FROM TIIE
SPECTRA O F CYCLOPENTADIENYL METALCOMPOVNDS DEPARTMENT O F CHEMISTRY, BROOKHAVEN NATIOXAL LABORATORY, HARVARD UNIVERSITY]
3689
I,ARORATORY, .4ND TIIE MALLIXCKRODI
Mass Spectra of Cyclopentadienyl Metal Compounds. Part I. Bis-cyclopentadienyl Compounds of V, Cr, Fe, Co, Ni, Re and Ru, and Manganese and Magnesium Cyclopentadienides BY L. FRIEDMAN, A. P. IRSAA N D G. WILKINSON RECEIVED FEBRUARY 5 . 1955 Mass spectra are given for the neutral bis-cyclopentadienyl compounds of T', Cr, Fe, S i , Co, R u and ( C ~ H S ) ~ R and ~H manganese arid magnesium cyclopentadienides. The stability of parent molecule ions is greater in the bis-cyclopentadienylmetal series Differences in the spectra of the latter and the manganese and magnesium compounds are correlated with the nature of the metal to ring bond in the respective compounds. Appearance potentials are given for the prominent ions ixi most of the compounds and additional support for the correlations made from the spectra is obtained from these data. Confirmation of reported isotope abundances of Y and Re also was obtained.
Discussion of Results Mass spectra of the (Cb&)&f derivatives of Mg, Mn, V, Cr, Fe, Co, Ni and Ru are presented in Table I with relative ion intensities normalized t o 100 units for the parent molecule ion. The spectra have been abbreviated to include only the prominent ions. An alternate representation of the data is found in Table I1 where major ions are tabulated as percentage of total ion yield. Total ion yields were computed omitting peaks of less than several percentage relative ion intensity. Sensitivity coefficients were not measured for lack of a suitable micromanometer or pressure gage in the and mm. range. Spectra of CloHl,&lg, CloHlaMn and CloHloReH are presented separately in Table 111, in somewhat greater detail, Experimental showing the alternate products of ionization and The methods of preparation of the following cyclopentadissociation by electron impact. dienyl metal compounds have been described previously I n all cases the highest mass ion obtained was the and samples of the original materials were used: CloHloFe,? molecule ion. No indication of association of moleCloHloCo,z CloHloSi,2 C I O H ~ O CC~l o, ~ H l ~ R u , ~C I O H I O Y , ~ cules in the vapor phase was detected and there was CloHloMn,6a CloH&.Ig,6b C~oHloRe.' The compounds of Fe, Co, Si,Cr, Mn and M g were no evidence of extensive thermal decomposition prepared from Mallinckrodt reagent grade chemicals. The in the ion source a t approximately 200". Most of rhenium was obtained from the University of Tennessee, the patterns were characteristic of those observed Knoxville. Term. The source of vanadium is not known to in -4XA type molecules such as Con, SO,, CS2, etc.% us. The (CsH5)nFe,S i , R u and KeH samples were purified Only minor amounts of dissociation involving loss by crystallization from petroleum ether and these and of C or H or hydrocarbon aggregates smaller than other specimens were sublirned three times in Y ~ C U Oa t 500 C5H5 from the ClOH1o-l.I unit were observed. The 10 . The mass spectra and appearance potentials were deter- spectra were, generally, unlike patterns obtained in mined with a General Electric analytical mass spectrometer studies on organometallic compounds* such as (SO' sector 6" radius single focusing). Since the compounds diethylzinc or dimethylmercury which undergo were not sufficiently volatile to use the conventional sample extensive dissociation of the parent ion. Some system and leak, samples were introduced directly into the moderate similarity between these spectra and ion source from tubes fitted with breakseals. The latter were necessary to permit preliminary evacuation of the those of CloHloMg and CloH1oMn exists but beyond spectrometer without loss or contamination of the sample. this the spectra are as different as those produced Sample pressures were controlled by gentle heating with from benzene and cyclohexane where in the former heating tape or the use of cooling baths. Ions were formed by electron bombardment with 50 volt electrons and spec- case the ring survives and dominates the spectrum tra were scanned magnetically a t ion acceleration potenwhile in the latter cleavage reactions predominate. tials of 1000 and 2000 volts. Previous attempts a t correlation of spectra of a family of moleculesg have emphasized two factors (1) Research supported in p a r t b y t h e U.S. .4tomic Energy Commission. in the relative probabilities of ionization and ( 2 ) G . Wilkinson. P. I,. Pauson and F. A C o t t o n , T H r s J O I . R N A L , dissociation processes. The first is that a necessary 76, 1970 (1954). condition for the observation of an ion is its stability ( 3 ) G. Wilkinson, i b X , 76, 209 (1954). over the time period required for its ejection from (4) G. Wilkinson, i b t d . , '74, 6146 (1952). f 5 ) E. 0 . Fischer . a n' d%1 H a f n e r , Z . Safurlovschg , 96, 503 (1954). the ion source and transit through the analyzer \Ye are indebted t o D r . Fischer for his kind gift of a specimen of tube t o the collector. The second factor deals CI~H,OV. with the energetics of the ionization and dissociation (6) ( a ) G. Wilkinson a n d F . .4. C o t t o n , Chemistry a n d I n d u s l v y (LonIntroduction Mass spectra of a series of neutral bis-cyclopentadienyl compounds of several transition metals and of magnesium and manganese cyclopentadienides were investigated with the objective of establishing a correlation of spectra with molecular structure. Sdditional interest in these metal compounds stemmed from their potential technical value as vehicles for the determination of isotopic abundances and masses of the respective metal nuclides. This paper contains mass spectra of the CloHlo.II compounds of Mg, V, Cr, Mn, Fe, Co, Ni, R u and CloHloReH and appearance potentials of the major ions in the spectra of all but the last two compounds mentioned.
--
11, 307 (1954); (b) C. U'ilkinson, J. Birmingham a n d F. A. Cotton, THISJ O U R K A L , in press. (7) G. Wilkinson, J . R I . Birmingham a n d F. A . C o t t o n , J . I n o r g . doli),
S i r c l e a r Chciir., in press.
(8) American Petroleum I n s t i t u t e Research Project 44. M a s s Spectral D a t a , Carnegie I n s t i t u t e of Technology, Pittsburgh, Pennsylvania. (9) L. Friedman a n d F. A. Long, THISJ O U R N A L , '76, 2832 (1953).
L.FRIEDMAN, -1. P.IRSAAND (3. LI-ILKINSOS
3690
TABLE I 1fASS
SPECTRA
I1 11
01: 1 f A J O R TONS R A S E D O S P A R E S T
Cr
1.
135.9 156.7
87,(i
lOll.0
100.0
5 5 .3 10.7 100.0
26.8
]:e
6 il
36. .i
34.3
73.5 100.0
24.0 1cin.0
RII
10tii
100
co
u,
25.2 20 . i) 18 3
12.5 40.7 46.8
28.2 20.8 31.0
16.6 22.3
18.7 20.8 (iiI . ,5
18.9
100
'1
14 7 lrj8,O
100 71.5
3.7 5'5.6 9.3 9 0
I
CsH:, 11 'CjHB
156
0
-
139.9 ,n 6.0 105" Posqible impurity or background contamir~atiori, 1)
process. Since most fragmentation takes place simple rupture of valence bonds, processes requiring least expenditure of energy are assumed to be most probable. Stability considerations are most important in connection with yields of niolecule ions which are assumed to be primary ionization products. Their respective yields are generallyinterpreted in terms of the probability of subsequent decornpositiori processes. The energetics of the respective processes is the major factor associated with the probability of production of atoniic metal ions since further decomposition is impossible i n these cases. Elementary considerations of the type noted above tend to oversimplify the complex processes occurring in the mass spectroineter ion source. I t is hoped that they may serve as a profitable basis for some brief discussion of the spectra. The niass spectra are not definitive for the deteriiiination of structures, but with the aid of supplementary structural data serve as a basis for comixirison of structures. Trends indicated by stahility of molecule ions are assumed to reflect the 5tability of parent compounds. The cyclopentatlienylriianganese antl magnesiurn derivatives are in a class by themseh-es. Their parent inulecule ions are relatively unstable contributing only 20Vo to the total ion yield in contrast to the Fe, Co, Cr, etc., conipouiids in which the inolecule ioii constitutes roughly .SOY2or more of the total ion yield. The instability of the molecule ions iti 1111and ,\fg compounds is reflected by enhanced yieldsof C5HjM+ a n d hl' ions. I t might be reasoned therefore that i n the C1,,HI,,llnand the CloHIclXSgcompounds, the
-!b
\'iI.:Lr)
I'C
6.5
171
l(lO.0
Cr
111
li I I
40.8
1.
, Relative ion intensity-ClnHlnlLIg CloHiaIln CloHluRe1-i
ClcHrlhI C,LlH\QM CioH~h1CsH8hlt C6H:L.I ' CsHeLZ CsHdhI C6HsR.I A CNahl' or
Ki
311 . 8
M 11
MASS SPECTRA O F CloI-IloMg, CloHloMii ASD CloHloReH Ion
10(1
27,l
T A B L E I1 AIASS SPECTRA BASEDo s T O T A L10s
TABLE
E
CI,
61.1
3.6 46.4
4 !i 18.%i 77.0
iiietal to ring l m i t l is of ciuite a different riaturc than the bond iri the other C-oHIJr compounds. This is supported by conclusions reached on other ground^.^ Thus in contrast t o what may be called the "sandwich bonded" molecules of which ferrocene, CloHloFe, is the prototype, in which the metal to ring bond is a two electron covalent bond of a rather special type, lo ClsHloXIti and CI(,H1oXIg have been shown to he essentially ionic compounds. This differelice in bond type is shown by notablc differences in chemical properties, by the formation of conducting solutions in liquid aiiinionia and by magnetic nieasurenients on the ~iianganeseconipound, which when in solution or magnetically dilute has a iiiagnetic moment corresponding to five unpaired electrons. -1 "sarid-rvich bonded" CloH14~Mn would be expcctedlO~l' to 1i:ive nnly one unpaired electroii. The similarity in the spectra of' CloHlliLIn antl CloHl,,?\lgas presented in Table I is corrected with ionic binding. Differences in spectra a t C H d I q ant1 CjHnf ions in the Mn atid LItl spectra may arise from slight sainple contaiiiiriation or background iri the mass spectroineter. It may be noted t h a t while extensive ionic character is attributed to the ring to metal bonds in CLIIHI,:SIIi and CNHlOMg, the niolecule ions CIIIHloSI are easilv detected, although as noted above their stability is less than t h a t of other molecular ions. For the V, Cr, Fe, Co, Xi, Ru aiid Re compounds, the metal to ring bond is unquestionably of the covalent "sandwich bond" type, and experimental evidence in support of this fact has been obtained from X-ray diffraction and magnetic measurements, infrared spectra, as well as from chemical properties.? -Attempts to obtain the mass spectrum of CloHIoTi have unfortunately failed. Difficulty in obtaining pure samples of this compound in the inass spectroineter ion source may be due to its instability on suhlinlatioll and probable t h e r i r d decomposition in the source. V'hile the mass spectra of CloHloFeand C M H ~ O C O xre closely similar, it is evident from Table I ancl I1 that significant differences exist between the various derivatives of the first transitional series. Cr, Xi aiitl V derivatives show slightly greater niolecule ion instability with an average ion yield close to 48';1, of the total in contrast to approxiniately 60voin Fe or Co. This is reflected in '101 J . D. Dunitz and I,. E . Orgei, V n t ~ ~ r 171, e. 121 ( I C h m i Plays., in pres3 L 1 1 ) 1%;. Moffitt, THIS J O I ' I I N A I . , 7 6 , .~3;38fii 1'254'
.I
SPECTRA OF CYCLOPENTADIENYL METALCOMPOUNDS
July20, 1955
3691
TABLE IV APPEARANCE POTENTIALS OF MAJORIONS(VALUESIN E.v.) hfg
M+ CsHhM’ CloHloM+ I(M 9
D ( CsHs-M) I (C5HsM)
14.36 f 0 . 2 10.98fO.l 7.76fO.l 7.64 3.36 7.62
Mn
V
Cr
14.05f0.2 11.25f0.2 7.25f0.1 7.43
18.32f0.2 12.68fO.l 7.56f0.1 6.74
16.15f0.3 13.6 f O . l 6.91f0.2 6.76
Fe
17.1 SzO.9 14.38f0.3 7.05f0.1 7.90
COMPUTED BONDENERGIES AND IONIZATION POTENTIALS 6.05 4.70 4.60 3.31 4.15 7.94 6.63 8.90 9.78 10.05
Table I by larger yields of the decomposition products C5H5M+ and M +. Thermochemical studies on ClcHloFe and CloHloNi2 support the argument that there is a weakening of the metal to ringbonds in CloHloNi amounting to 24 kcal./mole less than the corresponding bonds in ferrocene. Similar energy differences would be expected in the Cr and V compounds. Weakening of the metal to ring bond in the Cr, V and Ni derivatives does not suffice to account for differences observed in their spectra. Greater stability of CbH6V+ and C5H5Ni+ions must be postulated to account for the fact that the yields of V+ and N i + are equal t o or less than yields of F e + or Cof on a percentage of total basis, while in the C,oHloCr spectrum the C r + yield is 28% of the total or better than 50% greater than the corresponding processes in ferrocene, etc. I n CloHloCr weakening of Cr-C5Hs bond has the expected effect of increasing the yield of atomic ions a t the expense of CloHloCr+ and C5H5Cr+. I n Ni and V the CbH&I+ ions show increased stability over C6H5Cr+or C5H5Fe+ or C6H6Co+. Explanations of the relative stability of various bis-cyclopentadienyl-metal derivatives may be tested for consistency by their ability to account for stability of the C5H5M+ions. Increased yields of parent ion are observed in the spectra of CloHloRu and CloHloReH. Table I11 shows for the latter an extremely different type dissociation pattern with processes involving loss of an H atom from the parent as well as loss of CzH4groups replacing the “normal” loss of C5H5 radicals. Eoth Ru and Re derivatives give small yields of ill+ and CSH5M+ions along with large yields of parent ions, indicating structures probably more stable than CloHloFe or CloHloCo. Some evidence supporting a stronger metal to ring bond in ruthenocene is obtained from infrared spectra. l 2 I t may be noted finally that the compounds provide vehicles for the introduction of derivatives of the metal atoms into electron impact type ion sources facilitating the study of both isotopic abundances and atomic masses. The natural abundance of Ku isotopes already has been studiedl3 using CluHloRu as a source material. Data on other atomic systems were found to be in good agreement with published values. Of particular interest was the result with V where the isotopic abundance of V50 was checked in a completely different mass range and from probably different source materials. The ratio of VS0/Vs1was found to be 0.0028 f 0.0002 in good agreement with Hess (12) E . R Lippincott and R. D. Nelson, J. Chern. P h y s . , 21, 1307 (1953). (13) L. Friedman and A. P. Irsa, THISJ O U R N A L , 76, 5741 (1953).
Ni
co
16.1 f 0 . 6 14.20fO.l 6.2 f 0 . 3 7.86
14.32 f 0 . 2 12.67f0.1 7.06f0.1 7.63 3.35 9.32
and Inghram’s14result of 0.0026. This value is somewhat higher than the ratio 0.0023 obtained by Leland.15 The ratio of RelE5/RelB7 was found to be in good agreement with results of Cameron and White16 (0.59), who used Re207 as their source material. Appearance Potentials The vanishing current methodI7 was used to determine the appearance potentials listed in Table IV. Calibration of voltage scales and instrument sensitivity was made by running appearance potentials and ionization efficiency curves on krypton and neon. Aside from instrumental or experimental errors, the validity of the use of appearance potential measurements for the derivation of bond energies and ionization potential depends primarily on the correct designation of the process involved in the threshold measurement. I n this work the following processes are assumed for the production of the respective ions CICH&
+
+ e - --+ CioHioM’ + 2e-+ CnHsM+ + CjHs + 2e+ e - d Mi + 2CjHs
CloHloM eCioHioM
(1)
(2) (3)
with all products in their ground state and having no excess kinetic energy. I n reactions 1 and 2 errors arising from excitation of the products will cause observed values to be upper limits for the energetics of the designated processes. I n reaction 3 the possibility of formation of a CloHlo aggregate instead of two cyclopentadienyl radicals can lower the observed value below the true threshold for the process assumed, in addition to the potential error in the opposite direction brought about by formation of excited products. An additional difficulty comes from the fact t h a t calibrations involve “matching” the ionization efficiency curves of the respective product ions to curves obtained by ionization of rare gases. Stevensonla has observed that the shape of a family of ionization efficiencv curves varies with the complexity of chemical bond rupture processes involved and that processes involving cleavage of two or more bonds give rise t o longer tails or voltage differences between the linear rise of ion current with electron voltage and the threshold. The average tail length in the C,oHloM + ions is 1.4 e.v. and the value in the M + ions is 3.0 e.v. The deviations from the average tail lengths (14) D. C. Hess, J r . , and M. G. Inghram, Phys. Rev.. 16, 1717 (1949). (15) W. T. Leland, ibid., 76, 1722 (1949). (16) J. R. White and A. E. Cameron, ibid., 74, 991 (1948). (17) T. Mariner a n d W. Bleakney, ibid.,1 2 , 807 (1947). (18) D. P. Stevenson and J. H. Hipple, Jr., THISJ O U R N A L , 64, 1588 (1942).
of M A are about 0.5 e.v. with an unusual situation of 1.5 volt deviation in the Cr+ ion. Consequently the errors in the determination of the absolute value of appearance potentials are magnified by the difficulty in calibration. An alternative to the use of rare gas calibration is the assumption that, in C1?HloNi,for example, the sum of the ionization potential of Ni and the thermochemically determined bond dissociation energy is equal to the appearance potential of Xi'. This gives a value 0.7 e.v. lower than obtained using a K r calibration. This still gives poor internal consistency in the iron case with a n estimated bond energy higher than obtained from t h e thermochemical data. Results obtained for metal-ring bond energies by subtracting ionization potentials of the free metal atom from the appearance potential of the corresponding ion from C,oHloM indicates weaker bonding in CloH&g, CloHloMn and CloHloNi. The values for the compounds of Fe, Co, Cr and lrare higher but in view of the uncertainties these can be considered only as showing a trend indicating stronger bonding, Subtraction of the estimated bond energy from appearance potentials of C5H511+ ions gives an upper limit for the ionization potential
of the CsH5Xradical. If these potentials are compared with ionization potentials of the respective metal atoms a similar correlation is made separating ionization mechanisms for C5Hs3Ig) C5H5Mn and C5H5V radicals into one category with potentials 0 7 and 1.5 volts above the atomic ionization potentials and remaining ions with potentials ranging from 2.4 t o 3 2 . e.v. higher. Higher ioniation potentials indicate ionization mechanisms involving removal of electrons either from the C5H5ring or from ring-metal bonds rather than loss of electrons from non-bonding metal orbitals. Consideration of ionization mechanisms leads to the row in Table 111 containing molecular ionization potentials. In the Mg, Cr, M n and V coriipounds these values are all close to free metal ionization potentials. Differences of approximatelv 0.6 and 0.9 e.v. below atomic ionization potential are observed in Ni and Fe, respectively, with Co lower by the largest amount, 1.7 e:,-. The experimental data on molecular ionization potentials are probably least subject to error of all the appearance potentials Some qualitative arguments may be suggested for the low CloH1iCo+ value on the basis of a closed electronic configuration for the positive ion
[CONTRIBUTIOS FROM THE POLYTECHNIC I S S T I T C T E O F B R O O K L Y S ]
Chelation of Copper (TI) with Polyacrylic and Polymethacrylic Acid BY -1. 11. KOTLIAR~ AND H. NOR.IWETZ R E C E I V E D J.4SUARP
6,1%?12
The binding of copper( 11) ion by poly(acry1ic acid) and poly(methacr>-licacid) was itivestigated by compzriiig titratioii data, absorption spectra and dialysis equilbria. The extent of copper(I1) binding cannot be calculated from thc titratioii shift without assuming the nature of the chelate formed. Spectroscopic evidence indicates the persistence of a Tingle chelate, containing probably four carboxylates bound to a copper(I1) ion, over a wide range of conditions. Dialysis rquilibrium measurements can b e made with dilute polymer solutions and high pol>-rricr/copper(11) ratios and they yield a detailed picture of the dependence of chelation on electrostatic factors, polymer concentration arid the density of ligaiitl groups in t h e individual macromolecule.
The interaction of polyions with monovalent counter-ions has been thoroughly investigated in recent years and an excellent review of the work has been published by Doty and Ehrlich.? Much less is known about the behavior of polymeric acids in the presence of multivalent cations, where both electrostatic interactions and specific complex formation have t o be taken into account. A previous communication from this Laboratory3 dealt with the binding of alkaline earth ions by hydrolyzed maleic anhydride copolymers, a particularly simple case, since there is no uncertainty about the nature of the chelate complex. This paper deals with the chelation of copper(I1) by poly(acry1ic acid) (PALA)and poly(inethacry1ic acid) (PIZIX) . The procedure outlined I)reviously3 for the interpretation of the pH shift produced by a cation in terms of complex formatioil with 3 polymeric acid does nut yield unambiguous results if the number of ligands bound is unknown atid the titration &ita were therefore supplemented with :I ( 1 ) Abstrkcted from t h e 1R.i.5 Ph.11. theais of A. 11. K u L l i u , AI l o r a n e t z , A .\I. Kotlizir a n d H XIark, J I'hy.s ( ' h c n x 5 8 , 19 (1951).
direct determination of copper(I1) binding by dialysis equilibrium and spectroscopic studies of the nature of the chelate complex..' Results and Discussion LVhen (PMA) is titrated in 1 Npotassiuni nitrate, a pronounced shift t o lower p H values is observed on addition of small concentrations of copper(I1) ion (Table I). This was ascribed to coinplcx formation, since similar concentrations of all; a 1'l l l C earth ions produce no shift in the titration curve ( r i ' PAM-1in strong salt solution. T o interpret titration shifts brought about by :i chelating ion, the procedure introduced by Bjerruin' has t o be modified in accordance with polyelecirolyte theory. SiTe shall assume t h a t the electrical free energy o f dissociation LIE',,^ o f t h e carbos!.l groups oil the polynieric chain is independent oE tlic iiuiiiber o f bound metal ions as long as 3 , the avcrage uegative charge per carboxyl group is held con( 4 )
