GEO. GLOCKLER 35 IONIZATION AND DISSOCIATION BY ELECTRONIC IMPACT THE IONIZATION POTENTIALS AND ENERGIES OF SOME LIMITATIONS ON THE METHOD FORMATION OF SEC.-PROPYL AND TERT.-BUTYL RADICALS. BY D. P. STEVENSON Received 2znd January, 1951 It is shown that the mass spectrometrically measured appearance potentials of a variety of ions in a number of different substances are mutually consistent with the assumption that there are electron impact induced ionization and dissociation processes R, - R, + E -+ Rl+ 4- R, + ZE such that the appearance potential of R,, A(R,+), is given by the equality where I* ( ) and D( ) are the ionization potential and dissociation energy of the parenthetic substances, respectively. It is further shown that a necessary condition for the equality is I*(Rl) < I”(R,).If IS(Rl) > I*(R2), then i t is found that A(Rl+) > I*(R,) + D(Rl - R,) and probably the neutral product accompanying the formation of Rl+ is not R, but either R,* or F, + F, where the asterisk indicates electronic excitation and the F’s smaller dissociation fragments. The appearance potentials of C,H,,+, C3H,+, C,H,+ and C,H,+ in the mass spectra of a number of branched alkanes have been measured. It is found that in these cases that the condition, I*(R,) < P(R,), is satisfied, the appear- ance potentials of the ions C3H,+ and C,H,+ are mutually consistent and the A(R,+) = I”(R1) + wh- R,)36 IONIZATION POTENTIALS intercombination of the appearance potentials with the appropriate thermo- chemical data lead to D(sec.-C,H,-H) = 4-01 0.1 eV.D(tert.-C,H,-H) = 3-8, f 0.1 ,, P(sec.-C,H,) = 7*46 & 0.1 ,, P(tert.-C,H,) = 6.9, & 0-1 ,, It is thus found that the ionization potentials of the free radicals, CH,, C,H,, sec.-C,H, and tert.-C,H, do not parallel the ionization potentials of the cor- responding alkanes. It is suggested that i t would be of interest to determine whether the quantum theory offers an explanation of this lack of parallelism. It is shown that in certain mass spectrometers employing wolfram cathodes, there is insufficient differential pumping between the cathode and ionization chambers, and thus the apparent appearance potentials and intensities of certain ions in the mass spectra of some substances are falsified by the back diffusion of pyrolysis products in these mass spectrometers. The measurement of the so-called appearance potentials of the ions characteristic of the mass spectra of substances as excited by single electron impact in the dilute gas provides a potentially powerful means of studying the energetics of not only ions but also unstable molecules such as free radicals, The same difficulties are encountered in the interpretation of data obtained with mass spectrometers as are encountered in the inter- pretation of optical sFectra. These are the experimental ones of extra- polating observable intensities to limiting ones to determine character- istic energies and then associating the characteristic energies so determined with particular products and energy levels of the products.The mass spectrometric method has an advantage over optical methods in that in general the empirical formula of at least one of the products, the ionic one, is uniquely determined.This advantage is also a limitation on the method since it is thus required that only processes that lead to at least one ionic product can be studied. The experimental problem of extrapolating observations on ionization efficiency curves of ions in mass spectra to obtain the so-called appearance potentials that can be associated with definite energies has been solved by semi-empirical means. Methods of extrapolation have been found that not only yield reproducible limiting energies, the appearance potentials, but yield energies in agreement with those found by other methods. Thus extrapolating ionization efficiency curves for the molecule-ions in the mass spectra of a number of olefins in the manner found appropriate for agree- ment between the appearance potentials of rare gas ions and the spectro- scopically determined ionization potentials of the rare gases, leads to essential equality of the appearance potentials of the olefin molecule-ions and the ionization potentials of the olefins as determined from Rydberg series in their spectra in the vacuum ultra-violet., The interpretation of the appearance potentials of fragment ions in the mass spectra of complex substances involves the following problems.What are the neutral fragments that are simultaneously formed, and what are the states of electronic excitation of the neutral products and the ionic one ? I n addition to these problems the possibility arises that due to peculiarities of the potential hyper-surface of the state of the molecule-ion from which the fragments have been formed, i t has been necessary to endow the molecule-ion with more than the minimum energy necessary for the formation of the particular set of fragments in a par- ticular set of states. In the simplest case this would correspond to an activation energy for the formation of the molecule-ion, R,-R2+, from the radical R, and the ion R2+. It is the purpose of this discussion to show that in a number of cases the simplest set of assumptions suffice for the interpretation of the data to yield an apparently reliable determination of the energy of formation 1 Honig, J .Chem. Physics, 1948, 16, 105.D. P.STEVENSON 37 of the methyl radical. The methods employed for the determination of this energetic datum are employed to obtain the energies of formation and the ionization potentials of the sec.-propyl and tert.-butyl radicals from new experimental data on various Cs-cB alkanes. that the appearance potentials of certain ions in the mass spectra of propane and butanes could not be interpreted by means of a simple set of assumptions concerning their processes of forma- tion. Similar phenomena have been found with higher alkanes and a general rule is formulated for the prediction of those processes that may be interpretable through the simple assumptions. The origin of the failure of the simple assumptions has been explored by means of studies of isotopically labelled hydrocarbons.It has been found Experimental Measurements repoIted in this paper were made with a Westinghouse Type LV mass spectrometer somewhat modified in these laboratories. The modifi- cations and the method of measurement have been de~cribed.~~ The additive constant in the ionizing electron energy scale was determined by association of the initial break in the ionization efficiency curve of the argon ion, *OA+, with the spectroscopically determined ionization potential of argon, 15.76 eV.6 The possibility that the hydrocarbons might cause variation in the contact potentials associated with the electron gun was eliminated by making the measure- ments on argon simultaneously with those on the ions C,H5+, C,H6+, and C,H,+ in the mass spectra of the hydrocarbons.The contributions of the ion, C3H4+, to the m/q =. 40 current were corrected for by determining the relation between the ratios C,H4 +/C,H, + and C,H,+/C,E€, + as a function of apparent ionizing electron energy characteristic of the hydrocarbon mass spectrum in the absence of argon. The ratio C3H6+/C3H7+ was then used as a pseudo-energy scale for the determination of the contributions of C,H4+ to the m/q=40 positive ion current at any given apparent ionizing electron energy in the presence of argon. No detectable differ- ence was found between ionization effici- ency curves of argon as measured in the presence or absence of hydrocarbons. The appearance potentials were taken from the initial breaks of the ionization efficiency curves. The effects of range of the specific intensities of different ions were eliminated by the method of Smith,6 i.e.adjusting the intensity scale so that the linear portions of the ionization effici- IONIZING ELECTRON ENERGY-VOLTS FIG. I .--Ionization efficiency curves for the ions C4H,+ and C,H,+ of the 2 : 2 : 3-trimethylbutane mass spectrum. ency curve's were all of the same slope. The hydrocarbons upon which measurements were made were taken from the pure compound bank maintained by the Spectroscopic Department for calibration in spectrometric analyses. Linde Company spectroscopically pure argon was used. The appearance potentials that have been measured are given in Table I. Typical ionization efficiency curves are shown in Fig. 1-3. Stevenson and Hipple, J . Amer. Chem. Soc., 1942, 64, 1588.Stevenson, J . Chem. Physics, 1950, 18, 1347. Stevenson and Wagner, J . Chern. Physics, 1951, 19, 11. 5 Moore, Atomic Energy Levels (I Circular, Nut. Bur. Stand., c-467 (Washing- ton, D.C., 1949). 6 Smith, Physic. Rev, 1937, 51, 263.Molecule iso-C4Hl, . iso-C,HI2 . . z : 3-MeZ-Butane . z : 2-Me,-Butane . z : z : 3 : 3-Me4-Butane . neo-C,H,, . z : 2 : 3-Me3-Butane . FIG. 2.-Ionization efficiency curves for the ions C,H,+ and C,H5+ of the z : 2-dimethylbutane mass spectrum. A (CgHs') A (CSH6+) A (GH,') A(C4H8') A (CaH,') - 10.5~ f 0.1 11.0~ & 0.1 - - 13.2 0.2 10.2, f 0-1 10.8, f- 0-1 L - - 9.8, & 0'2 I0'Ys & 0'1 - - 13.6 & 0.2 I - 9'3 & 0'2 I0'Is & 0'1 - - - 10.32 f. 0.1 10.2~ & 0.1 I - 11.8, & 0 - 2 9'52 & 0-2 10'Oo 3 0'1 - - 9-24 0'1 9'79 & 0'1 - .O IONIZING ELECTRON ENERGY-VOLTS FIG. 3.-Ionization efficiency curves for the ions C3H,+ and C,Ho+ of the isopentane mass spectrum.Discussion The simplest assumption that can be made with respect t o the signifi- cance of the appearance potential of an alkyl ion in the mass spectrum of an alkane is that it corresponds to the energy of the process, - R, - R, + -+ Rl(X) + R,(X) + zE A (Rlf) = 'z(R,) f D(R1 - R Z ) J where X indicates the ground electronic state, A ( ) the appearance poten- tial, Iz() the ionization potential and D ( ) the dissociation energy. If such conditions obtain for the formation of an ion in the mass spectra of two substances, then the combination of the appearance potentials of the ion with the appropriate thermochemical data permits the determina- tion of a dissociation energy. For example, the appearance potentials of the ethyl ion in the mass spectra of ethane and propane in combination with the heats of formation of methane, ethane and propane, and theD.P. STEVENSON 39 dissociation energy of hydrogen into hydrogen atoms would yield the dis- sociation energy of methane, D(CH,-H), if C,He + Z? -+ C,H5+(X) + H(,.S+) + LF A(C2H5+) = Iz(C2H5) + D(C,H,--H), C,H8 + E +C2H5+(X) + CH,(X) + 2E The dissociation energy of methane, D(CH,-H), computed in this manner (termed the indirect method) could be in error if the two processes did not yield the ethyl ion in the same state, if the methyl radical were formed in an excited state or if either or both processes, and A (C,H5+) = IZ(CzH5) + D(CZHS-CH3).H + C,H5+ + CzH,+ CH, + C,H5+ + C3H8+ require an activation energy. It has been found that the appearance potentials of molecule-ions themselves can be reasonably accurately associated with the ionization potcntial of the molecule. Thus, the dissociation energy of methane, D(CH,-H), can be alternately calculated as the difference between the appearance potential of the methyl ion in the methane mass spectrum and the appearance (ionization) potential of the methyl ion in the methyl radical mass spectrum. Such a determination of D(CH,-H) would of necessity be greater than or equal to the true value, and it would be greater only bv virtue of an activation energy for the process H + CH,+ --f CH,+. This second method of obtaining dissociation energies from appearance potentials is termed " the direct method ".Agreement between deter- minations by the two methods would be strong argument for the validity of the simple assumption concerning these electron impact induced pro- cesses stated above. In the case of D(CH,-H), five pairs of processes have been examined and found to yield values that agree well within their experimental error. Furthermore, a direct measurement of the ionization potential of the methyl radical combined with either of two appearance potentials of the methyl ion gives D(CH,-H) in excellent agreement with the values deduced from the appearance potential pairs by the indirect method. These data and results are summarized in Table 11. There are also shown in this Table the results of determination of D(CH3-H) from the kinetics of photochemical reactions and pyrolyses. The complete agreement between the various determinations of D(CH,-H) shown in Table I1 can be taken as evidence that electron impact induced processes of the simple type described above do exist.It further suggests that the data in Table I may be similarly interpreted to yield dissociation energies of propane and isobutane, D (sec.-C,H,-H) and D (tert.-C,H,_H), respectively. The combination of the appearance potential, A(C,H,+) in the mass spectra of isobutane, isopentane and z : 3-dimethylbutane with the auxiliary data, z : 3-Me2-Rutane + CH, = iso-C,H,, + C,H,, = 0.18 eV,, z : 3-Me2-Butane -+ C,H, = iso-C,H,, + C,H,, AH:,, = 0.4 ell,, CH, = CH, + H C2H, = C,H, $- H D == 4-42 eV.D = 4.20 eV.8 7 Rossini et al., Selected Values of Properties of Hydrocarbons (Circular of the Nut. Bur. Stand., c-461, Washington D.C., 1947). See (a) of Table 11.40 IONIZATION POTENTIALS TABLE II.-TTARIOUS DETERMINATIONS OF D (CH,-H) Method Direct E.I. . Indirect E.I. . Electron Impact Average . Photochemical . Pyrolysis , ~ Process (a) CH, 3 CH,+ + (b) CH, -+ CH,+ + H + ( c ) CH,OH --f CH,+ + OH + E (d) C,H, -+ C,H6+ + H + E with C3H8 C,Hs+ + CH, + E (e) C,H8 -+ C,H7+ + H + 3 with 1 isoC,rI,, -+ C,H7+ + CH, + (f) C,H, 3 C,H6+ + H + with 7 isuC,H, + C,H,+ + CH, + (g) n-C,H7C1 -+ C,H7+ + C1 + with 7 (h) CH,OH 3 CH,OH+ + H + E with C,HsOH -+ CH,OH+ + CH, + E with I n-C4H,o + C3H7+ + CH3 + E J 3 (i) CH, + Br 3 CH, + HBr HBr -+ H + Br H I + H + I ( j ) CH,I -+ CH, + 1 D(CHI, -- H) eV (a) Hipple and Stevenson, Physic.Rev., 1943, 63, IZI. (b) Ref. (6) of text. (c) Cumming and Bleakney, Physic. Rev., 1940, 58, 787. ( d ) Stevenson, J. Chem. Physics, 1942, 10, 291. (g) Stevenson and Hipple, J . Amer. Chem. SOG., 1942, 61, 2766. (h) See (c). (i) Kistiakowsky et al., J . Chem. Physics, 1942, 10, 305 and 653 ; 1943, 1 1 , 6. ( j ) Polanyi et al., Nature. 1940, 146, 129 and 685 ; 1941, 147, 542 ; Trans. Famday SOC., 1941, 37, 377 and 648 ; 1943, 39, 19. result in D(sec.-C,H,-H) = 4-01 f 0-1 eV, Iz(sec.-C,H,) = 7*46 f 0-1 eV, while from A(C,H,+) in the mass spectra of neopentane, z : z-dimethyl- butane, z : z : 3-trimethylbutane and z : z : 3 : 3-tetramethylbutane and the auxiliary data z : z : 3 : 3-Me4-Butane + CH4= neo-C,H,, + iso-C,H,,, AH:,, = 0.03 eV,? z : 2 : 3 : 3-Me,-Butaiie+C2H,=z : 2-Me,-Butane+iso-C,H,,, AH:,, z : z : 3 : 3-Me,-Butane+C3H,=2 : z : 3-Me,-Butane+i~o-C~H,~, AH:,, and D(CH3-H) and D(C2H6-H) - -0*07,7 - 0-07, - D(tert.-C,H,-H) = 3.8, -& 0-2 eV, Iz(tert.-C,H,) = 6.9, & 0.2 eV, as summarized in Table 111.These dissociation energies for secondary and tertiary C-H bonds are in approximate agreement with the values given by Butler and Polan yi, D(sec.-C,H,-H) = 3.86 eV and D (tert.-C,H,-H) = 3-73 eV.D. P. STEVENSON 41 Process TABLE III.-VARIOUS INDIRECT ELECTRON IMPACT DETERMINATIONS OF D (sec.-C3H,-H) AND D(tert.-C,H,-H), AND OF IZ(sec.-C,H,) AND I* (tert.-C,H,) D(R-H), I*( R) ~ s o - C ~ H ~ O -+ C3H7+ + CH3 + E - } 2 : 3-Mez-Butane --f C3H,+ + C,H, + E iso-CsHlz -+ C3H7+ + C,H, + - } 2 : 3-Mez-Butane --r C,H7+ + C3H, + E 4.Oz 4'Il 7'54 - neo-C,H,, .+ C4Hs+ + CH, + E 2 : 2 : 3 : s-Me,-Butane --f C4H9+ + C4H9 + E } 2 : 2-Me-Butane 3 C4H9+ + C,Hs + z : z : 3 : 3-Me,-Butane --f C,H,+ + C,H, + E -} 7'36 3-89 3437 av.R = sec.-C,H, I 4-07 f 0.1 7'46 z t 0'1 av. R = tert.-C,H, 2 : 2 : 3-Me3-Butane + C,H,+ + C3H7 + 2 : 2 : 3 : 3-Me4-Butane -+ C4H9+ + C,H, + E "} 3.8, & 0.1 6-90 Jr 0-1 D(sec.-C3H7-H)-D(tert.-C,H9-H)=0~2 vs . 4-07 - 3.8,=0.1, 6.8, 6.9, The electron impact values for the dissociation energy of alkyl-hydrogen These bonds differ in several respects from those from pyrolysis kinetics. differences are : (i) a smaller range between D(CH,-H) and D(tert.-C,H,-H) according to the electron impact method, i.e.0-54 eV compared with 0.72 eV and (ii) the electron impact method gives less difference between D(C,H,-H) and D(sec.-C,H,-H) and greater difference between D(sec.-C,H,-H) and D(tert.-C,H,-H), respectively, than does the pyro- lysis method. According to Steacie the activation energies for the react ions CH3 + C2H6 +CH4 + CZHS CH, + C3H8 --f CH4 + C3H7 CH3 + iso-C4H1,3 + CH4 + C4H9 are 8, 6-8 and 4-2 kcal./mole, respectively. To the extent that these activation energies reflect the strengths of the primary, secondary and tertiary carbon-hydrogen bonds, they suggest the order of the inequalities, D(C,H,-H) > D(sec.-C,H,-H) > D(tert.-C,H,-H), by the electron im- pact method t o be preferable t o that given by the alkyl iodide pyrolysis method.It may be noted in this connection that the difference between the ionization potentials of propane and isobutane (0-9-1.0 eV) lo is greater than the difference between the ethane and propane ionization potentials (0.4 eV).% l o The ionization potentials of the alkyl radicals, CH,, C,Hs, sec.-C,H, and tert.-C,H,, IO-O,, 8.6,: 7*4s and 6-9, eV, respectively, roughly parallel those of the corresponding alkanes, namely, 13.3,~ 1.1-7,z 11.3 lo and I 0-4 2 eV, respectively. However, the free radical ionization potentials do not show the discontinuity of trend between C, and C, shown by the alkane ionization potentials. It is not obvious to the author whether or not exact parallelism would be expected to exist between the ionization potentials of these two homologus series on theoretical grounds.New York, 1946), p. 520. 9 Steacie, Atomic and Free Radical Reactions (Reinhold Publishing Corp., lo Delfosse and Bleakney, Physic. Rev., 1939, 56, 256,42 IONIZATION POTENTIALS Comparison of the mass spectrum of ordinary propane with those of n-propyl deuteride and sec.-propyl deuteride 4 has conclusively shown that in the formation of C,H,+ in the propane mass spectrum it is one of the secondary hydrogens that dissociates. The sum of the ionization potential of the sec.-propyl radical and the dissociation energy of the secondary carbon-hydrogen bond given above is 11.5~ & 0.1 eV. This agrees within experimental error with the value found for the appearance potential of C,H7+ in the propane mass spectrum, 11.67 & 0.1 eV.” It has also been shown that in the formation of C4H0+ from isobutane, it is the tertiary hydrogen that dissociates.The sum of the ionization and dissociation energies, Is(tert.-C,H,) and D (tert.-C,H,-H), given above, is 10.8 f 0.3 eV. This is considerably less than the value found for the appearance potential of C4H9+ in the isobutane mass spectrum, I 1-57 fo.2 eV.2 Since there are no sufficiently low states of the hydrogen atom, the difference between the calculated and observed appearance potential must be associated with excitation of the butyl ion. If the previously found appearance potential of C4Hg+ in the mass spectrum of tert.-butyl chloride, 10.2, & 0.1 eV,12 is combined with the heat of formation of tert.-butyl chloride (- 1-83 eV) and the dissociation energy of HCl (4.43 eV),13 the energy required for the formation of C4Hs+ + H from isobutane is calculatcd to be 10.8, & 0-1 eV, in excellent agreement with the value from the sum of ionization potential, I*(tert.-C,H,), and of dissociation energy, D(tert.-C,H,-H), given above.Since the appearance potential of C4Hg+ in the tert.-butyl chloride mass spectrum is thus shown to be compatible with the data reported in this discussion, the dissociation energy of the tertiary carbon-chlorine bond may be taken as the difference between this appearance potential and the ionization potential of the tert.-butyl radical, D(tert.-C4Hs-C1) = 10’2, - 6.9, = 3 ~ 3 ~ eV. It may be noted that neither the appearance potential of C2H,+ in the isopentane or z : 2-dimethylbutane mass spectra nor that of C,H,+ in the z : 2 : 3-trimethylbutane mass spectrum has been employed to deduce dissociation energies.The reason for these omissions will now be discussed. It was found that the appearance potentials of the methyl ion in the mass spectra of propane and the butanes 2 were considerably greater (ca. 5-8 eV) than the minimum values to be expected from the sum of the ionization potential of the methyl radical (10.0, eV) and the dissociation energy of the carbon-carbon bonds being broken, if the processes of forma- tion were C,Hs + .i? --f CH,+ + C,H,(X) + zE, - Thus, we may conclude the processes are not represented by the equations a s written, although the seemingly complementary processes giving ethyl and propyl ions, respectively, are represented by the equations C,Hs + E -+ CH,(X) + C,H,+ + 2E C,H,, + or C4HIo + E + CH,+ + C,H,(X) + zE.- --f CH,(X) + C,H, + 2E Examination of the mass spectra of the monodeutero-propanes and butanes reveals the processes yielding CH,+ must be quite different from those yielding C,H,+ (propane) or C,H7+ (butanes). The intensities of the ions C,H,D+ in the mass spectra of the isotopic propanes, n-C,H,D and sec.-C,H,D are approximately those to be expected from their molec- ular structure, and similarly for the intensities of the lions C,H,D+ in the mass spectra of the monodeutero-butanes.4 However, contrary to 11 Stevenson and Hipple, J . Amer. Chem. Soc., 1942, 64, 2769. 12 Ref. (g) of Table 11. 1, Herzberg, Molecular Spectra and Molecular Structure (Prentice Hall, hTew 1-ork, 1939).Closer scrutiny of the mass spectra of the various isotopic propanes and butanes suggests a possible reason for require- ment of an excited state of the molecule-ions, C,H,+ and C,H,,+ for the formation of CH,+.Al- 3 though the intensities of C,H,+ g and C,H,D+ in the mass spectra ; of n-propyl deuteride, and of 2 C,H,+ and C,H,D+ in the mass 5 spectra of the butyl deuterides were approximately those to be L~J expected from the molecular structures, it was found that B and similarly for n-butyl deuter- ide.4 Since such a large fraction ,.oo 5 0.80- o.40- n-C3H7D C,H,D+z 1.2 C,H,+ 0.20- of the propane and butane mole- 00- - - - 20.0 40.0 60.0 80.044 IONIZATION POTENTIALS and while and It is seen that extra energy of the magnitude of 1.2-1-7 eV is required.It may be of significance for the nature of the process that in these cases the extra energy is approximately that required for reactions of the type These results suggest the phenomenon first observed in the formation of the methyl ion in the mass spectra of propane and the butanes is indeed a general one and that there exists the general rule that for IS(R,) > 18(R,), the process R, - R, + E -+ R,+ + R, + ZE Iz(C,H,) - Ia(tert.-C,H,) = 8.67 - 6.9, = 1 . 7 ~ eV, triptane, A (C,H,+) - A (CIHO+) = 1-7, eV, Is(sec.-C,H,) - Is(tert.-C,H,) = 0.5, eV. alkyl radical = olefin + H. A H = 1.6 - 1.8 eV. requires A(%+) > WR,) + D(R,-R,), while for the complementary process R,-R, + E -+ R, + R,+ + 2 E The existence of this rule greatly limits the potential of the electron impact method of determining dissociation energies.Having reliable values for the ionization potentials of the methyl and ethyl radicals, it was hoped that measurements of the appearance potentials of either CH,+ or C,H,+ in the mass spectra of CH,-R or C,H,-R would suffice to determine the energy of formation of the radical R by the formulae D(CH,-R) = A (CH,+) - Is(CH,) or D(C,H,-R) = A (C,H,+)-18(C,H,). However, for this direct procedure to be applicable, it is necessary that P(R) > I"(CH,) or I'(R) > Is(C,H,). These conditions make the method inapplicable to all C, and higher alkyl radicals. The appearance potentials of the olefin ions, C3Ha-f- and C4H8+, in the mass spectra of the iso-alkanes, Table I, present an interesting example of a difficulty that may beset electron impact studies.In agreement with previous findings 2 the appearance potential of C,Ha+ in the mass spectrum of isobutane, 10.5 f 0-1 eV, equals the sum of the ionization potential of propylene 1 (9.80 & 0.05 eV) and the heat of the reaction, iso-C4H,, = C,H, + CHI, AH& = 0-80 eV. However, this appearance potential in the isopentane mass spectrum is lower than the least energy for the formation of C,H,+ from isopentane, iso-C,H,, + E -+ C3H6+ + C,Ha + zE, A(C,H,+) > 9.8 + 0.92 = 10.7 eV, and this appearance potential ( A (C,Ha+)) in the 2 : 3-dimethylbutane mass spectrum equals the ionization potential of propylene 1 It can only be concluded that isopentane to a small extent and z : 3-dimethylbutane to a greater extent pyrolyze at the wolfram cathode to give propylene among other substances, and that this propylene diffuses back into the ionization chamber and thus falsifies measurements of C ,HI+.The appearance potential of C4H8+ in the neopentane mass spectrum, A(C,H,+) = 10.3 0.1 eV is equal to the sum of the ionization potential of isobutylene (9.35 eV) 1 and the heat of the reaction, neo-C,H,, = iso-C,H, + CH,, AH,,, = 0.80 eV. For z : 2-dimethylbutane, 2 : 2 : 3- trimethylbutane and z : 2 : 3 : 3-tetramethylbutane, the appearance potential of C,H,+ is essentially equal to the ionization potential of isobutylene. Hence, here, too, pyrolysis and back diffusion must occur.* It does not seem likely that these pyrolytic reactions can lead to pro- ducts that would cause error in the measurements of the appearance potentials of the alkyl ions.This belief is based on the observation that * I t should be noted that these appearance potentials of C4H8+ strongly support Honig's determination of the ionization potential of isobutylene and his conclusion that the value found by Stevenson and Hipple ( J . Amer. Chem. Sot., 1942. 64, 2769), 8.9 eV, is in error. A (%+) > Is(%) + D(R,-R,).D. P. STEVENSON 45 with decreasing molccular weight the appearance potentials of alliyl ions increase. The low pressures obtaining in the mass spectrometer preclude association reactions that could lead to higher molecular weight substances which might cause low apparent appearance potentials. Furthermore, such substances would have been detected in the recordings of the com- plete mass spectra.If the pyrolyses lead to appreciable quantities of alkyl radicals C,H, and C4Hs, and these are sufficiently long lived to diffuse through the four slits that separate thk cathode from the ionization chamber, it is conceivable that the appearance potentials of C,H,+ and C4Hs+ given in Table I1 are low in the cases other than isobutane and neopentane. However, the internal consistency of the appearance poten- tials indicates such effect, if any, must be less than the experimental error. It is to be noted that the pyrolysis and back diffusion is not unique to the Westinghouse Type L.V. mass spectrometer. Koffel and Lad,'" using a spectrometer of quite different construction, have reported for isopentane, A (C,H,+) = A (C,H,+) - 0.9 eV ; for 2 : 3-dimethylbutane, A(C,H,+) = A(C,H,+ ) - 1.7 eV ; for 2 : a-dimethylbutane, A(C4H8+) = A (C4H,+) - 1.0 eV ; and for 2 : 2 : 3-trimethylbutane, A (C,H,+) = A(C4Hs+) - 1.1 eV.These differences are quite like those reported in Table 11, and thus we conclude their mass spectrometer also suffers from insufficient differential pumping between the cathode chamber and the ionization chamber. In view of the above, it is apparent that the reported intensities of such ions as C,H,+, C4H,+ and the like in the mass spectra of branched alkanes, the A.P.I. 44 Catalog of Mass Spectra, are of no real significance as far as indicating the probability of the formation of such ions by electron impact induced dissociations. The fact that the appearance potentials of the propylene and butylene ions in the mass spectra of isobutane and neopentane, respectively, are essentially equal to the sum of the heat of the reaction ; alkane = olefin plus methane and the ionization potential of the olefin indicates the absence of significant activation energy for the reverse reaction, olefin ion plus methane equals alkane ion. The absence of activation energy for this reaction indicates there may well be no activation energy for a re- action of the type, n-alkane ion -tsec.-alkyl ion plus sec.-alkyl radical. Thus it seems likely that the electron impact method is inapplicable to the evaluation of energies of formation and ionization potentials of C, and higher normal alkyls. Incomplete interpretation of data on the appear- ance potentials of C,H,+ and C4H,+ in the mass spectra of n-pentane, n-hexane, n-heptane and n-octane appears to substantiate this conclusi~n.~~ The author wishes to express his deep appreciation of the late Dr. Otto Beeck's continued encouragement in the here-described and other research. Shell Develo$mend Company, California. Emeryville, 1" Koffel and Lad, J . Chem. Physics, 1948, 16, 420. These authors made an extensive investigation of electron impact induced processes in the C,-C, alkanes. However, their use of linear intercepts as measures of appearance potentials renders their data non-comparable with the present work. Further, it has been shown that (Physic. Rev., 1943, 63, 121 : J . Cham. Physics, 1950, 18, 1347) appearance potentials deduced in this manner (linear intercepts) are not the minimum energies for dissociation processes. However, the linear intercepts do provide an approximate measure of the difference between appearance potentials. "That such ambiguity as this might arise in the interpretation of alky ion appearance potentials has been suggested t o the author by Prof. S. Winstein, private communication.