J. Phys. Chem. 1991, 95,8344-8351
8344
Electronic Structure Considerations for Methane Activation by Third-Row Transition-Metal Ions Karl K. Irikura and J. L. Beauchamp* Arthur Amos Noyes Laboratory of Chemical Physics.' California Institute of Technology, Pasadena, California 91I25 (Received: February 21, 1991)
Methane is spontaneously dehydrogenated in the gas phase by many metal ions of the 5d transition series. In most cases, the MCH2+produced undergoes further reactions, leading eventually to products such as WC8Hl6+.The reactivity of the third-row transition-metalions, both ban and with simple ligands, may be explained in termsof electronicstructure considerations. Promotion energy, exchange energy, and the relative and absolute sizes of the valence s and d orbitals all appear to be important.
Introduction Most studies of the gas-phase chemistry of transition-metal ions have dealt with first-row metals.' Although alkane dehydrogenation and demethanation are common in these systems, no examples of facile, exothermic methane activation were known for many years. Endothermic reactions of the first-row ions with methane (reactions 1-3), however, have been well-studied by using M+ CH4 MCH2' + Hz (1)
+
-
+
+ CH3 MCH3' + H
MH+
(2)
(3) guided-ion-beam techniques.z In the first row, reaction 1 is observed only for the early transition metals S c + 4 r + . Both the electron configuration and the spin state of the metal ion are important for reactivity. For example, electronically excited Cr+ dehydrogenates methane far more efficiently than translationally excited Cr+ in its electronic ground Recently, studies have been extended to the second- and third-row transition metals, and many of these ions have been found to dehydrogenate methane exothermically. Although Zr+ is the only second-row metal found to react? many of the third-row Os+, Ir+, and Pt+)are very active in this regard, metals (Ta+, W+, often reacting sequentially (reaction 4).68 The doubly-charged +
M+ + nCH4
-
MC,H2,+
+ nH2
D(M+-C,H%) 1 17.8n + AH&Hzn) kcal/mol
(4) (5)
ions Zr2+, Nb2+, and Ta2+ have recently been discovered to be even more reactive than their singly-charged counterparts with methane and other hydrocarbon~.~*~J~ If reaction 4 is exothermic, a lower bound on the metal-ligand bond strength may be calculated by using equation 5." It is not immediately apparent that the principles of reactivity and bonding applicable for first-row transition-metal ions will also apply in the third row. In particular, strong spin-orbit coupling in the heavy metals weakens the validity of basic concepts such as total spin and electron configuration.12 If the rules for the first row do indeed apply in the third row,we can infer that the fundamental chemical concepts remain intact despite the strong spin-orbit coupling. Unligated transition-metal systems provide prototypes for basic processes in organometallic chemistry. For example, eq 4 is strikingly similar to the putative methylene coupling step in the Fischer-Tropsch synthesis of hydrocarbons (FTs).13 Methylene coupling in the FTs is generally assumed to require multiple metal atoms at an appropriate surface site.I4 In the gas phase, methane oligomerization occurs at a single metal center, suggesting that a single atom may suffice in the heterogeneous FTS as well. In other related chemistry, a highly selective Ca/Ni/K oxide catalyst for the oxidative coupling of methane has recently been rep01ted.l~ 'Contribution no. 8400.
0022-3654191/2095-8344S02.50/0
Unlike previous coupling catalysts,16 the new chemistry appears to involve only surface reactions. One reasonable mechanism would involve a metal-carbene intermediate, as in eq 4, followed by desorption of the C2 product.
Experimental Section Experiments were conducted by using a Fourier-transformion cyclotron resonance (FTICR) spectrometer equipped with an electromagnet,operated at 2.0 T, and an IonSpec data system." Pressures were measured by using a Schulz-Phelps ionization gauge18 calibrated against an MKS Baratron capacitance manometer (model 390 HA-0001). Gas pressures in the range 10"-1O4 Torr were maintained by using variable leak valves (Varian). Pulsed gases were introduced through two General Valve pulsed valves (series 9) in tandem. Standard double-resonance techniques, such as ion ejection, collision-induced dissociation, and collisional activation, were performed when appropriate. These procedures involve excitation at or near cyclotron resonance in order to increase an ion's kinetic energy. Time plots, the standard procedure for kinetics studies, involve monitoring the complete mass spectrum as a function of time. Ions are generated by ablation of metal targets with the focwed output (308 nm) of a Lumonics excimer laser, using typically (1) Russell, D. H., Ed. Gas Phuse Inorgunic Chemistry; Plenum: New York, 1989. (2) Armentrout, P. B.; Beauchamp, J. L. Ace. Chem. Res. 1989, 22, 31 5-321. (3) (a) Freas, R. B.; Ridge, D. P. J . Am. Chem. Soc. 1980, 102, 7129-7131. (b) Halle, L. F.; Armentrout, P. B.; Beauchamp, J. L. J. Am. Chem. Soc. 1981, 103,962-963. (4) Georgia&, R.; Armentrout, P. B. J. Phys. Chem. 1988,92,7067-7074. ( 5 ) MacMahon, T. J.; Ranasinghe, Y. A.; Freiser, B. S. J . Phys. Chem., submitted for publication. (6) Irikura, K. K.;Beauchamp, J. L. J. Am. Chem. Soc. 1989, I l l , 75-85. (7) Buckner, S.W.; MacMahon, T. J.; Byrd, G. D.; Freiser, B. S.Inorg. Chem. 1989, 28, 3511-3518. (8) Irikura, K. K.;Beauchamp, J. L. J. Am. Chem. Soc. 1991, 113, 2769-2710. (9) (a) Buckner, S. W.; Freiser, B. S. J. Am. Chem. Soc. 1987, 109, 1247-1248. (b) Gord, J. R.; Freiser, B. S.; Buckner, S.W. J. Chem. Phys. 1989, 91, 7530-7536. (10) Ti2+only forms an adduct with methane: Tonkyn, R.; Weisshaar. J. C. J . Am. Chem.Soc. 1986, 108, 7128-7130. (1 1) Unless noted, auxiliary thermochemical data are from: Lias, S.G.; Bartmess, J. E.; Liebman, J. F.; Holmes, J. L.; Levin, R. D.; Mallard, W. 0. J . Phys. Chem. Ref. Dura, Suppl. 1988, 17, no. 1. (12) Experimental atomic data are from: Moore, C. E. Aromic Energy Levels. NSRDS-NBS 35 (reprint of NBS circular 467); U. S. Government Printing Ofice: Washington, DC, 1971; Vol. 3. (13) Anderson. R. B. The Fischer-Tropsh Synthesis; Academic: Orlando, FL, 1984. (14) Masters, C. Ado. Orgunomel. Chem. 1979, 17.61-103. ( I S ) Pereira, P.; Lee, S.H.; Somorjai, G. A.; Heinemann, H. Curul. LLtf. 1990.6, 255-262. (16) (a) Lunsford, J. H.; Cisneros, M. D.; Hinson, P. G.; Tong, Y. D. Furuduy Discuss. Chem. Soc., 1989, 13-21. (b) Tong, Y. D.; Lunsford, J. H. J. Chem. Soc., Chem. Commun. 1990,792-793. (17) Ionspec Corp., 17951 Skypark Circle, Imine, CA 92714-6323. (18) Schulz. G. J.; Phelp, A. V. Rev.Sci. Insrrum. 1957,28, 1051-1054.
0 1991 American Chemical Society
Methane Activation by Third-Row Transition-Metal Ions TABLE I: Rates of Dchydrogemtion of Methane0 ion CH, CDIb ion 3.4 2.4 ReCH2+ Ta+ TaCH2+ 2.0 0.6 ReC3H4+ 2.0 1.4 Os+ TaC2H4+ TaC,H6+ 1.4 1.3 OsO+ 1.2 * Hf, Ta, and W) with oxygen-containing impurities such as H 2 0 and O2are problematic, especially at high pressures or long reaction times. Some of the reactions are very sensitive to the kinetic energy of the reactant ion, so that the usual ion isolation procedures lead to nonreproducible rate constants or nonexponential decay. In such cases, ion isolation and the concomitant translational heating must be avoided, and rate constants are best determined by least-squares fitting of data to kinetic models.22
1.5
0 2
W+
COZ CO2
Reo+ Reo2+ ReC H ReC:Ht+ ReC4H4+ ReC4H6+ Re+
0 2
02
Reo2+ Reo3+
Ir+ Ir+ Ir+
C2H6 C2H6
[ReC6H+6+] 1r c2H 2 IrC2H4
CH30H
e
0 2
wo+ +
*Rates in units of cm3 s-l, estimated to be accurate to *25%. cm3 s-l cm3 s-l for CHI and 8.8 X Collision rates are 9.8 X for CD4: Gioumousis, G.;Stevenson, D. P. J . Chem. Phys. 1958, 29, 294-299. bRate in the fully deuterated system. 'Reference 6. dProduct is OsOCH2+. eIncludes -20% OsC2H4+product. /After apparent relaxation that occurs at the same rate. #For initially formed IrC2H4+,which appears to become inert at the same rate (see text).
wo+ wo2+ wo+ wo2+
wo+
C2H6 C2H6 C2H6 C2H6 C3H8
1.o
d d d d +
0.6 0.2 2.2 0.2 2.9 2.4 2.0 0.9 0.028 4.4 7.1 18
kAMb 5.6 17.7 17.7 17.7 17.6 9.8 5.6 5.6 6.7 6.7 5.6 5.6 9.7 9.7 9.6 9.6 9.8 9.7 9.7 14.7
cm3 s-'. Rates are estimated to be accurate ORates in units of to &25%. bCollision rates: Su, T.; Bowers, M. T. Inf. J . Muss Spectrom. Ion Processes 1973,12, 347. 'HfO+ and HfOH+ in 1.6:lratio. d Undetermined dehydrogenation products. e Reaction sequence not determined; see text.
-
Ethane undergoes single- and double dehydrogenation to HfC2H4+and HfC2H2+(in a 2:l ratio). These reactions imply D(Hf+-C2H4) > 32.6 kcal/mol and D(Hf+-C2H2) > 74.6 kcal/mol. CH3CD3yields HfC2H4+, HfC2HsD+,and HfC2H2D2+ in the statistical 1:3:1 ratio. Translational excitation of the H P reactant results in a product isotope ratio of 1 5 1 . HMD+ but not HfD2+is formed in an endothermic reaction. HM2+can not be definitively excluded because of its mass coincidencewith HfD+, another endothermic product. Formaldehyde yields an increase in the HfO+ signal as well as a peak due to HfH2+. This dihydride undergoes sequential H/D exchange with added D2: HM2+ + 2D2
+
HfD2+ + 2HD
(6)
Results Tantalum The reactions of Ta+ with hydrocarbons have already been reporteda7 Four sequential reactions with methane lead to Rate constants for reactions of atomic metal ions and orTaC4H8+. Rate constants were not reported in the earlier work; ganometallic fragments with methane are summarized in Table the values from our study are included in Table I. Reactions with I. Rate constants for reactions involving other neutral molecules residual air lead to products containing oxygen, including are listed in Table 11. TaOC,H2,+ ( n = 0-4) and Ta02C2H4+. Lanthanum. La+ was not included in this study because its reactivity with hydrocarbons has already been i n ~ e s t i g a t e d . ~ ' ~ ~ ~ Results of our collision-induced dissociation (CID) studies of TaCH2+, TaC2H4+,and TaC4H8+are in accord with those reThe only available third-row D(M+-CH2) value is D(La+ 74.6 kcal/mol). Likewise, reaction of W+ with acetone yields mainly WC3H20+, WOCH2+, and possibly WO+, but no W(CH2)2+. These assignments were confirmed by using acetone-d6. There is kinetic evidence for more than one isomer of wc4H8+. After it forms rapidly in methane, half of the lesWC4H8+(242 amu) reacts to give a species at 244 amu (probably WOC3H6+), which in turn reacts completely to give a peak at 246 amu (probably W02C2H4+).The other half of the WC4H8+is unreactive. This partial reactivity is observed even when the IMWC,H8+is not subjected to isolation pulses, so it is not the result of inadvertent collisional activation. Likewise, preparing Is6WC,H8+from isolated I"WC3H6+ results in the same kinetic behavior. We note that the dual reactivity is easily explained if half of the peak at 242 amu is due to WC3H40+,which might be formed by reaction of WC3H6+with OF Labeling experiments using I3CH4would resolve this ambiguity. In connection with the observation of oxygenated species, reactions with H20, 02,and COz were briefly studied. WO+ is the initial product in all cases;the rate constants that were measured are included in Table 11. While attempting to displace ethylene from WC2H4+,we found that C2D4 reacts sequentially with W+ to lead eventually to WCloDlo+.This is similar to the behavior of Nb+, which reacts with ethylene six times to yield NbCI2Hl2+ and also involves extensive ligand coupling.' Rhenium. Like H P , Re+ reacts endothermically with methane. The product can also be reduced with H2 back to the atomic ion (reaction 8). ReCH2+reacts quickly with methane to produce ReCH2+ + H2 ReCHz+ + CH,
-
-+
Re+ + CH4
(8)
ReC2H4++ H2
(9)
ReC2H4+(reaction 9). Similarly, although Re+ does not react spontaneously with ethane, the endothermic products ReC2H2+ and ReC2H4+react sequentially. Reaction of Re+ with propane is very slow, implying either a reaction barrier or slight endothermicity. The initial product is difficult to identify because it reacts at a much greater rate than that at which it is formed.
.
. -.. ..
.-
.. .. .. m ... ,* Metnane Activation PY I nira-Kow I ransition-Metal ions .I
The Journal of Physical Chemistry, Vol. 95, No. 21, 1991 8347
,
I
ReC6H6+is the most prominent and persistent product in both the ethane and propane sequences. Re+ is unreactive with 02, but Reo+ reacts twice to give Reo3+. Since the failure of Re+ to react with methane is thermodynamic rather than kinetic, it is reasonable to suppose the same for the failure to react with 0 2 Cyclopropane reacts with Re+ to give ReCH2+ (as well as ReC3H2+and ReC3H4+). Ethylene oxide is also reactive, with products including both Reo+ and R&H2+. There is an apparent increase in Re+ reaction rate with time. In addition, the isotope isolation procedure imparts sufficient kinetic energy to the selected ion (such as 18'Re+) to cause a substantial drop in the reaction rate. We consider the peculiar kinetics to be indicative of a reaction rate that decreases dramatically with increasing kinetic energy. The reactions described above imply D(Re+-CH2) = 102 f 9 kcal/mol and also D(Re+-O) = 102 f 17 kcal/mol (with the upper limit somewhat tentative). The lower bounds D(Re0'4) > 1 19 kcal/mol and D ( R e O 2 + 4 ) > 119 kcal/mol are also established. Evidence for partial oxidation of methane was sought in mixtures of methane and oxygen. Collisional activation is necessary to generate the endothermic product Reo+, which reacts spontaneously with CH, to form ReOCH2+,in analogy to OsO+.6 Another method for generating R e 0 2 ions is by electron impact on the vapors above heated Re03,25 Reo2+thus formed reacts with methane twice to give Re02C2H4+(reaction 10). This is Reo2+ + 2CH4
-
+
ReO2C2H4+ 2H2
+
CH4
k
(IrCH;)*
k
(r_ IrCH2'
k2
IrCaHe'
(IrCzH,')*
Model: ....'. minimal excited IrCH2+
-
Reaction Time (ms) 1 .o
Model: .....- excited Irl3CD2+ excited Irl3C2Dq+ also
-
(10)
unlike the case of OS02+ which , undergoes doublebond metathesis to OsOCH2+ followed either by a second metathesis step (to OSC2H4') or simple dehydrogenation (to OsOC2H4+). Reo2+ and Os02+ are similar in their reactions with ammonia; sequential metathesis occurs to form the corresponding MN2H2+species. Osmium. The reactions of OsO,+ (n = 0-4) with methane have been studied previously;6rate constants are included in Table I. The rate measured for the reaction with methane in the present work is in agreement with the earlier value. Iridium. Ir+ reacts three times with methane to give IrC3H6+. Subsequent reactions involve residual oxygen and generate IrC2H40+,IrC3H20+,and IrC4H40+(as confirmed by using CD,). Time plots of the reaction of IrCH2+with methane show an apparent increase in rate with time, as in some of the Re+ reactions mentioned above. This behavior persists even when the I S H 2 +is not subjected to isolation pulses, which could affect its translational energy. As illustrated in Figure 3 (Top), a good fit to the data can be obtained by using a kinetic model in which the initially formed IrCH2+ must undergo a relaxation step before it can react further.26 The rate of reaction of IrCH2+ (and to a lesser extent, of Ir+) decreases with increasing ion kinetic energy. As a result, rates often appear lower when measured with a single isotope of iridium instead of the natural combination of isotopes. We also note that the reactivity of IrCH2+ with methane is enhanced when O2 is added as a buffer gas. At long reaction times, the reaction of IrC2H4+with methane appears to stop. As shown in Figure 3 (Bottom), this kinetic behavior can be modeled by using a scheme in which the initially-formed IrC2H4+is slowly converted to an unreactive form: Ir'
1 .o
(11) IrC2Hd'
kzr
Reactions with ethane were briefly investigated. Sequential single- and double dehydrogenation are the principal processes
0
100
200
300
500
400
Reaction Time (ms)
- - -
Figure 3. Reaction of 19%+ with methane. (Top) With 1.3 X lod Torr of CH4. (-.) Least-squares fit using the simple kinetic model Ir+
- - -
IrC2H4+ subsequent products. (-) Fit assuming initially unreactive ISH2+;that is, Ir+ (IrCH2+)* IrCH2+ IrC2H4+ subsequent products. (Bottom) With 3.0 X lo" Torr of '%De fit to excited Ir13CD2+model as above. (-) Fit to a model that includes relaxation of IrI3C2D4+to an unreactive form. IrCH2+
(e..)
(implying D(Ir+-C2H4)> 32.6 kcal/mol and D(Ir+-C2H3 > 74.6 kcal/mol), although there are minor products that appear to contain odd numbers of carbon atoms. In contrast to the abundant formation of ReC6H6+from Re+, no IrC6H6' is evident in the Ir+ system. Instead, IrC6H8+and IrC6HI0+dominate at long reaction times. In connection with the much-sought direct conversion of methane to methanol, reactions with methanol were examined briefly. As with methane, there is an apparent increase in rate over time when a single iridium isotope is selected, but natural abundance Ir+ yields clean, pseudo-first-order kinetics. The products are K O + ,IrCH20+,and a small amount of IrH2+. No IrO+ or IrCH2+is formed. Exothermic formation of the dihydride requires that the average Ir-H bond strength be at least 63 kcal/mol. If IrH2+is a dihydrogen complex, then D(Ir+-H2) > 22 kcal/mol?' Substantial IrCH30H+ is also formed, presumably by displacement of the CO ligand in IrCO+ by CH30H. Dehydrogenation and ligand substitution continue to predominate in subsequent reaction steps. At long reaction times, the spectrum is dominated by a peak corresponding to IrC4H804+. Platinum. Pt+ reacts with methane sequentially to give species as high as PtC5Hlo'. Only the first step is efficient. An attempt to produce. PtO+ by collisional activation in O2was unsuccessful. Gold. An extensive study of the gas-phase chemistry of Au+ has been published.28 Although Au+ will slowly dehydrogenate ethane, it is unreactive with methane. In the present work, Au+
~~~
(25) Work done in collaboration with Edmund H. Fowles. (26) In every fit, the rate of (IrCHz+)* relaxation has been found to be equal to the rate of the subsequent reaction of relaxed IrCH2+. The significance of this coincidence is unclear at this time, but may indicate an oversimplified kinetic model.
(27) For comparison, there is an estimate D(Cp'Cr(CO),-H,) = 16.7 1.2 kcal/mol: Howdle, S.M.; Healy, M. A.; Poliakoff, M. 1.Am. Chem. Soc. 1990, 112,4804-4813. (28) Chowdhury. A. K.;Wilkins, C. L. 1.Am. Chem. Sac. 1987, 109, 5336-5343.
8348 The Journal of Physical Chemistry, Vol. 95, No. 21, 199'1
Irikura and Beauchamp SCHEME I
120
n
110-
I
I
T
.*&.
I W
+)+I
4-
W
n m
.
0 A
50 20
30
40
BDE(M+-H)
50
1st row 2nd row 3rd row
60
70
(kcal/mol)
Figure 4. Bond strengths D(M+CH2)and D(M+-H) (from Table VI) for (e) first-row, (0)second-row, and (A)third-row transition metals.
was inadvertently generated by laser ablation of a gold mesh that covered the laser target. In accord with the prior work, no reaction was observed with methane. A lower limit D(Au+-CH2) > 95 kcal/mol has been established.28 Mermry. Hg+ was not included in this study. It is not expected to react with methane because its d'Os' electron configuration cannot form a double bond.
Discussion In contrast to the general inertness of first- and second-row transition-metal ions with methane, many of the third-row ions react readily. In the third row, La+, H P , Re+, and Au+ do not react with methane, and although W+ reacts with CH,, it is unreactive with I3CD4. Ligands can enhance the reactivity. For example, ReCH and Reo+ do react with CHI, and W"CD2+ does react with 3CD4. One would therefore wish to explain why the third-row ions are reactive, why some of them are not reactive, and how ligands can enhance the reactivity of the metal center. Note that although D(M+-CH2) and D(M+-H) are correlated, as illustrated in Figure 4, there is no clear relationship with predictive value. Orbital Size. Metal-ligand bonds are generally stronger for third-row transition metals than for those of the first and second rows.29 As a result, the third-row metals are expected to be more reactive than their lighter congeners. The lanthanide contraction and relativistic effects lead to changes in orbital size and stability, which in turn result in increased bonding overlap and less loss of exchange energy upon b ~ n d i n g . ~ *The ~ ~ relative sizes of the valence s and d orbitals are strongly affected by the lanthanide contraction. The valence d orbitals show the expected trend going down the periodic table, with sizes increasing in the order 3d < 4d C 5d. The s orbitals, however, do not obey this trend, and the The 6s and 5d orbitals are therefore sizes are 4s = 6s C relatively close in size for third-row metals ( R , - Rd = 0.65). As a result, s-d hybridization is more effective. The s-orbital sizes are summarized in Table 111. As usual, orbitals contract as one proceeds from left to right in the periodic table. Another consideration is that the metal-carbon u and ?r bonds are expected to have optimal bond lengths that depend upon the sizes of the metal s and d orbitals. If the s orbital is much larger than the d orbital, then the optimum length for the u bond, which will be much greater than that contains substantial s ~haracter,~' for the ?r bond, and the metal-carbon distance will represent a compromise between u and a bonding. Since the sizes of the s +
(29) Martinho Simhs, J. A.; Beauchamp, J. L. Chem. Rev. 1990, 90, 629488. (30) (a) Pyykka, P. Chem. Rev. 1988, 88, 563-594. (b) Ziegler, T.; Snijders, J. G.; Baerends, E. J. In The Challenge of d and f Elecfrons; Salahub. D. R.;Zemer, M.C., Eds.; ACS Symposium Series 394; American Chemical Society: Washington, DC, 1989; Chapter 23. (31) Ohanessian,G.; Brusich, M. J.; Goddard, W. A., 111. J . Am. Chem. SOC.1990, 112,7179-7189. (32) Ohanessian, G.; Goddard, W. A., 111. Acc. Chem. Res. 1990, 23, 386-392. (33) Schilling, J. B.; Beauchamp, J. L. Organometallics 1988.7, 194-199.
TABLE III: Promotion Energies,Excbauge Energy Lases, OrMCll Sizes, and Reactivity witb Methane
prom.
tot.
ion
E"
exchb
Ec
r,d
La+ Hf+ Ta+ W+ Re+
4.0 10.4
4 18.1 30.6 43.9
8 28.5 30.6 43.9 58.0 41.5 31.3 31.9 47.9
2.37 1.91 1.80 1.73 1.69 1.64
Os+ Ir'
Pt+ A d
0 0 0 0
0 13.7 43.0
58.0 41.5 31.3 18.2 4.9
1.61 1.59 1.56
&/&,*
D(M+-CH,)f
0.0 0.0 0.3 0.1 0.0 0.3 0.7 0.4 0.0
98.2 2 102 9
*
>I11 >I11 102
9
>111 >I11 >I11 >95
Energy required to reach the d%' configuration (kcal/mol). bExchange energy lost upon forming a double bond with s and d orbitals (kcal/mol; ref 39). CPromotionenergy plus exchange energy lost. dRoot-mean-squareradius of the 6s orbital (ref 31). #Efficiency of reaction with methane (from Table I). 'Bond strength from Table VI (kcal/mol). and d orbitals are best-matched for the third-row metals, both metal-carbon bonds can be accommodated with relatively little compromise. The high charge in transition metal dications causes a substantial contraction of the valence orbitals, so that the d orbital is of similar size to the open-shell orbitals of the CH2 fragment.34 The increased charge also stabilizes the valence d orbitals relative to the s (because of their lower principal quantum number), so that the u bond in M=CHZ2+ probably contains little metal s character. Hence, the early metals (dn configuration) can form two strong bonds of nearly equal optimum length in MCH22+ (Scheme I). This may be the explanation for spectacular bond strengths such as D(Nd2+-CH2) = 197 f 10 kcal/mol?a Note that the double charge also leads to a deeper well for the initial ion-molecule complex, so that more energy is available for overcoming reaction barriers. Promotion Energies. Variations in energies of bonds to transition-metal ions have been rationalized successfully in terms of the energy required to promote the metal ion from its ground state For MH+ to a state that is well-suited for bonding.29*31*32*3s*36 species, the appropriate state often involves the d W configuration, but in the second and third transiton series the dnconfiguration is sometimes more relevant. The actual bonding will always involve both s and d orbitals to varying degree^."*^'*^^ We find that the reactivity of the third-row transition-metal ions with methane is well-explained by using a model in which the d%I configuration is considered to be required for bonding to a methylene fragment. Each of the third-row metals is discussed below in terms of this model. J-averaged atomic energy levels are ordinarily used for comparison with nonrelativistic ab initio c a l ~ u l a t i o n s . ~ 'The ~~~,~~ (34) Irikura, K. K.;Goddard, W. A,, 111. Unpublished results. (35) Carter, E. A.; Goddard, W. A., 111. J. Phys. Chem. 1988, 92, 5679-5683. (36) Armentrout, P. B.; Sunderlin, L. S.;Fisher, E. R.Inorg. Chem. 1989, 78 441f+-4411 --, . .- - , .- . . (37) Schilling, J. B.; Goddard, W. A,, 111; Beauchamp, J. L. J . Phys. Chem. 1987, 91, 5616-5623. (38) Schilling, J. B.; Goddard, W. A., 111; Beauchamp, J. L. J. Am. Chem. SOC.1987, 109, 5565-5573.
Methane Activation by Third-Row Transition-Metal Ions TABLE I V Romotioa Energies, Exchange Energy Losses,and Orbital Shea for Croup 6 Metal Ions prom. tot. exchb Ea ion Ee RP 4-w 1.73h 1.03h 37.e 71.8 Cr+ 34.g 72.0 1.87) 0.8Y 33.7' 34.21 Mo+ 43.9 1.73'" 0.61'" 0' 43.9' W+
Energy required to reach the d%' configuration (kcal/mol). bExchange energy lost upon forming a double bond with s and d orbitals (kcal/mol). 'Promotion energy plus exchange energy lost. dRoot-mean-squareradius (in A) of the valence s orbital. 'Difference in rms radius of valence s and d orbitals. 'Sugar, J.; Corliss, C. J . Phys. Chem. Ref. Data, Suppl. 1985, 14, no. 2. 8Reference 35. "eference 37. 'Sugar, J.; Musgrove, A. J. Phys. Chem. Ref. Datu 1988, 17, 155-239. /Reference 38. kReference 12. 'Reference 39. Reference 3 1. promotion energies quoted here are instead those between the lowest J levels in the two terms involved. For the first two transition series there is little practical difference, but spin-orbit coupling is so strong in the third row that different terms often overlap substantially. For example, the levels of the ground state (d2,3F) of La+ are at 0,2.9, and 5.7 kcal/mol. The first excited state (d's', ID),lies among the ground state levels at 4.0 kcal/ mol.I2 For the present purposes, it is therefore more appropriate to refer to the lowest level in each term. La+ is not subject to the lanthanide contraction, so its valence orbitals are unusually large and d i f f u ~ e ,reducing ~' the bonding overlap with the CH2 moiety. In addition, as noted above, 4.0 kcal/mol is required to attain the favorable dlsl configuration. Two of the other three unreactive ions ( H P and Au+) also require promotion energy to reach the d%' configuration, as summarized in Table 111. Moreover, the reactive ions Ta+-Ir+ have de's1 ground states appropriate for bonding. Pt+ is also reactive, although it requires 13.7kcal/mol in promotion energy. As a late metal, its orbitals are evidently small enough (Table 111) to compensate for this requirement. Exchange Energy. Although Re+ has a dssl ground-state configuration suitable for bonding, it is unreactive because of the large amount of exchange energy (the energy behind Hund's rule) lost upon bonding.32 Exchange energy losses are a maximum for Re+, which has five s-d interactions (Kd) and 10 d-d interactions ( K d ) . Bonding to the 6s (u) and a 5d ( r )orbital results in loss of approximately 58 kcal/mol (2.5Kd + 2Kdd).39 In contrast, ReCH2+loses only about 36 kcal/mol (2.5Kdd)upon bonding to a second carbene fragment, and does indeed react (efficiency = 0.2). Likewise, although W+ loses 44 kcal/mol in exchange energy upon bonding to CH2, WCH2+loses only 20 kcal/mol, accounting for its greater reactivity. Exchange energies are affected by orbital size. Moving down the periodic table, the valence d orbitals grow larger and the sizes of the s and d orbitals tend to converge.32 Since the exchange interaction depends inversely on the distance between electrons, Kdddecreases and Kd increases from the first to the third row. In a number of first-row transiton-metal complexes M(H20),,+, the second water ligand is bound more strongly than the first, apparently because the first ligand "pays" the promotion energy of the metal center.40 For third-row ions MC,,Hb+, the second carbene appears to be more strongly bound because the first ligand "pays" for the loss of exchange energy. Perhaps the third factor, orbital size, is similarly affected in some ML, species; the first ligand may reduce the size of the metal orbitals (perhaps by an increase in oxidation state) and thus prepare the metal for bonding to another ligand.
The Journal of Physical Chemistry, Vol. 95, No. 21, 1991 8349 TABLE V: Ligand B M h g Energies Required4
structure +M=CH2
D(M+-€.Hd 111
49
structure CH,
CH,
ll-+M-ll CH, cH,-
D(M+C.H,.) 97
CH, +
M>>
146
69 CH,& H-
CH,
+
M)
145
'0
203
+ M 3
+M
).>
138(
59
H-+M))-
153
'In kcal/mol; calculated by using eq 5. bEstimating pHr(1,3propanediyl) = 2AHdI-C3H7)- AHI(C3H8). 'Estimating AHd1,4butanediyl) = 2AHl(n-C4H9)- AHr(C4Hlo).
(39) Values of Kd and KM for the first- and second-row M+ are available in ref 35. Values for Re+ are in ref 31. We estimate the exchange energies for the remainder of the third row by using the following simple scaling formula: K(3rd) = K(2nd) + AK(2nd - Ist)[AK(Re+ - Tc+)/AK(Tc+-
In summary, this straightforward analysis serves to explain the pattern of reactivity with methane among third-row transitonmetal ions. The important factors include (1) promotion energy, (2) exchange energy loss, and (3) orbital size. Table I11 provides a quantitative summary of these factors, as well as kinetic and thermodynamic data for the reactions with methane. Comparison among Transition Series. The reactivities of the group 6 metal ions Cr+, Mo+, and W+, provide a good illustration of the principles discussed above. Promotion energies, exchange energies, and orbital sizes are listed in Table IV. Although exchange energy losses are comparable, W+ requires no promotion energy. In addition, it has by far the least mismatch between sand d-orbital sizes, and is correspondingly the most reactive. In contrast, Mo+ reacts only slowly with hydrocarbons, and Cr+, with the worst mismatch in orbital sizes, does not react at all." The electronic structure considerations affect product distributions as well as overall reaction rates. First-row metals react with hydrocarbons to give products of both C-H and C-C bond cleavage. In the second row, less carbon loss (such as demethanation) occurs, and dehydrogenationis more common and e x t e n s i ~ e . ~ ~Because * ~ ' * ~ ~of this trend, the powerful dehydrogenating ability of the third-row transition-metal cations is not unexpected. Thermochemistry alone cannot explain this preference for dehydrogenation. For example, demethanation of ethane by Ta+, W+, Os+, Ir+, or Zr+ is exothermic by at least 15 kcal/mol, yet only dehydrogenation is observed. The chemistry is thought instead to be under kinetic control, with competition between 8-hydrogen migration (leading to Hz loss) and &alkyl migration (leading to alkane loss).'2 The relevant orbital of a migrating alkyl group is much more directional than the spherical orbital of a migrating hydrogen atom. If the metal orbital is also directional, then it will be difficult to maintain overlap with the alkyl orbital during the migration process. In contrast, overlap with the hydrogen atomic orbital is easily maintained. The second- and third-row metals employ greater d character in their bonds.32*37.38 They are therefore expected to favor hydrogen migration and consequent loss of H242 The greater d character is probably due to the better size match between the
(40) (a) Magnera, T.F.; David, D. E.;Michl, J. J. Am. Chem.Soc. 1989, 1 1 1 , 4 1 ~ 1 0 1 .(b) Marinelli, P.J.; Squires, R. R. J. Am. Chcm. Sm. 1989, 111.4101-4103. (c) Rosi, M.; Bauschlicher, C. W., Jr. J. Chcm. Phys. 1990, 92, 1876-1878.
(41) Byrd, G. D.; Freiser, B. S. J. Am. Chcm. Soc. 1982,104,5944-5950. (42) Tolbert, M. A.; Mandich, M. L.; Halle, L. F.; Beauchamp, J. L. J. Am. Chem. Sac. 1984, 106.5675.
Mn+)].
Irikura and Beauchamp
8350 The Journal of Physical Chemistry, Vol. 95, No. 21, 1991 SCHEME IV
SCHEME I1 H-H
H---H 8
,
/” +H&CHz W(+CH,
-
-
+Hf=CH,
/H
“*
+Hf,
H, +Hf =CH2
H’
SCHEME 111 HfCH,CD,+
H
+Hf’ ‘CH2CD3
-
+ HD
-+Hf
I
H,C=CD,
HfHD’
Both C2D4 and CH$N fail to displace ethylene from WC4Hs+.
/D
-
In addition, CID of TaC4Hs+leads to loss of neither ethylene nor CH2? These results argue against a bis(ethy1ene) structure. The scrambled producu
‘CH,CD,H
+ CH,CD,
(endothemc )
s and d orbitals in the second- and third-row metals, which is expected to lead to more effective s-d hybridization. As supplemental evidence for greater d orbital character in bonding, we note that the rotational barrier around a metal-ethylene bond is substantially greater (more metallacyclopropane character) for iridium than for rhodium.“ Structure rad Meebanism. The exothermicity of reaction 4 implies very strong binding of the h y h r b y l ligand to the metal (eq 5).” For example, a bis(ethy1ene) (AHH,= 25 kcal/mol) structure requires the metal-olefin bonds to exceed an average of 48 kcal/mol. Several possible structures are evaluated in Table V. While the required metal-ligand bond strengths arc all high, they are not sufficiently unreasonable to permit any structures to be excluded on thermodynamic grounds. We suppose that oxidative addition is the initial step in all of the reactions with methane. This is consistent with the strong deuteriuum isotope effect in the W+ reaction, and also was a s u c c d u l working hypothesis in our earlier work with ox”ium cations? As has previously been argued for the reaction between La+ and CH4,U the oxidative addition of methane to HtY is probably followed by concerted [2, + 2.1 elimination of H2 (Scheme 11). The hafnium(V) (methylene) dihydride intermediate is dismissed as involving an unreasonable oxidation state of hafnium. We postulate an analogous concerted mechanism for the H/D exchange of H M 2 + with D2. These Mbbius-type electrocyclic reactions are symmetry-allowed when one of the bonds involves a d The labeling pattern among the products of the reaction of HfC with CH3CD3suggests the competition illustrated in Scheme 111. Initial oxidative addition is followed by a &hydrogen migration. The ethylene dihydride intermediate can either dissociate to products or revert to the ethyl hydride, leading to scrambling. As the internal energy of the complex is increased,dissociation appears to k a m e more rapid and predominates. When the energy is suffident to release the endothermic HfHf product, no scrambling occurs at all. The results of the mossover experiment involving Ta+ indicate that the second reaction with methane occurs by one or both of the paths outlined in Scheme IV. Path a also leads to the TaC2H4+product, but could lead to crossover only by an unusual reversible a-methyl migration.
lack of H/D exchange with C2D4 or D2 implies that there are no hydride ligands in WC4Hs+. The similar reactivity and CID patterns for the WC4Hs+ions generated from methane and from cyclopentanone are consistent with similar structures. Since Ni+ reacts with cyclopentanone to give a nickel (acyclopentane) ion,& it is reasonable to suppose that W+ also forms a metallacycle. We therefore favor a metallacyclic structure for the MC4Hs+ions from the sequential reactions with methane. This structure requires an average M-C bond strength of at least 69 kcal/mol (Table VI. As discussed above, both IrCH2+ and IrC2H4+appear to be subject to relaxation processes that affect their reactivity. Possible forms of initial excitation include translational, electronic, vibrational, and chemical (isomerization) energy. Translational excitation does indeed decrease the rate of reaction of IrCH2+. It is however unreasonable to expect significant kinetic energy to be retained in the primary and secondary products, when none is apparent in the initial Ir’. Furthermore, the iridium-containing ions are much more massive than methane, so that little of the kinetic energy would actually be available in the center of mass. Electronic excitation is plausible if IrCH2+ and IrC2H4+ are formed in unreactive spin-excited states, since changes of spin could easily be slow enough to be observable on the ICR time scale. Radiative and collisional relaxation of a vibrationally hot product would probably also be slow. Vibrationally hot products, however, suggest a very tight transition state, which is probably incompatible with the high kinetic efficiency. An isomerization is unlikely; it is difficult to construct reasonable pairs of isomers, to explain why unstable isomers would be formed, and to describe how they would subsequently isomerize. The increase in reactivity of IrCH2+upon adding O2buffer gas does not distinguish among the Merent types of intenal energy, since triplet O2is expected to be an effective collision partner for translational, vibrational, and spin relaxation. Additional experiments are needed to identify the origin of the unusual kinetic behavior. NbH+ produced in the reaction of Nb2+ with ethane has also been shown to be internally excited and to undergo endothermic reactions.% In this case, translational excitation is quite reasonable, since the products are both heavy cations and repel each other strongly in the exit channel. M e t h e Oxidation. There is long-standing interest in oxidative coupling and direct partial oxidation of methane in order to make better use of natural gas?’ We consider two relevant reactions of methane, oxo-metal double-bond metathesis and the direct conversion of methane to methanol by a metal oxide (reactions 12 and 13). The gas phase systems may be able to provide simple M-O
+ CH4
M=O
+ CH4
+
-
(43) (a) Arthurs, M.A.; Nelson, S. M. J. Cowd. Chem. 1983,13,29-40.
(b) Artburs, M.; AI-Daffaec, H. K.; Haslop, J.; Kubal, G.; Peareon, M. D.; Thatcher, P. J. Chem. Soc., Dalron Trans. 1987, 2615-2619. (44) Steigerwald, M. L.; Goddard, W. A., 111. J. Am. Chem. Soc. 1984, 106, 308-31 1. (45) Rappt, A. K. Organome~allics1987.6, 354-357.
+
M=CH2 H20 AH = D(M-O) - D(M-CH2) - 6.6 kcal/mol (12) M
+ CHiOH
AH = D(M-O) - 90.0 kcal/mol (13)
(46)Jacobson, D. B.; Frciscr, B.S.J. Am. Chem. Soc. 1983,105,736-742. (47) (a) J. S.;Oyama, S. T. Coral. Rea-Sci. Eng. 1988, 30, 249-280. (b) Pltchai, R.; Klier, K. Curd. Reu.-Sci. Eng. 1986, 28, 13-88.
e,
Methane Activation by Third-Row Transition-Metal Ions
M sc Ti V
D(M+-CH# 99 5
* 94 4
80 i 3 54 4 71 3 83 4 78 2 75 2 64 2 95 3
Cr Mn Fe
*
co Ni
cu Y Zr
>Illd
Nb Mo Tc
106
* 4b*
D(M+-H)b 57 f 2
D(M+-OY
54 3 48.5 f 1.5 32.5 2 48 A 3 50 & 1.5 46.5 1.5 40 i 2 22 3 62.5 1.5 55 3 54 3 42 3 463'
165 134 77 59 75 66 49 33 180 208 208 114
* ** *
In kcal mol. From ref 29 unless noted. tentative. Reference 6. 'Reference 28.
/
1st row
2nd row
The Journal of Physical Chemistry, Vol. 95, No. 21, 1991 8351
M Ru Rh Pd Ag La Hf Ta W Re
os Ir Pt
Au
From data in ref 1 1. dReference 5. 3rd row
D(M+CH2)b
D(M+-H)b
91 * 5 98 1.5 102 P >Ill>Ill# 102 f 9r >Ill#* >Ill# >Ill# >9Y e Reference
94 56 50 208 174 189 127 102
loo
173 12'
60 60
7. f Reference 32. #This work. *Upper limit
The sequence of reactions initiated by Ir+ is interesting in this context. A number of oxygenated ions are formed in mixtures of methane and oxygen, as described in reactions 15 and 16. Since IrCjH6' 0 2 IrCjH20' HzO H2 (15 )
+
+
+
+
+
IrC3H20' + CH4 IrC4H40' Hz (16) the bond in IrO+ is weak (60kcal/mol), the oxygen in the products is probably not an oxo ligand; C-O bonds are probably formed instead. These products are worthy of further study to determine if the ligands can be removed as interesting products. +
-100-
D(M+-O)C
* *
41 3 36 3 47 f 3 16 3 58 f 2 54.9, 54.6 49.9, 44.9 56.6 65.8, 62.9, 33.4
sc
Mn
cu
Transition Metal
Figure 5. Energetics of metal-oxo and metal-methylene bonds. (0)and ( 0 )represent bond strengths D(M+-O) and D(M+CH2), respectively. ( 0 )indicates the calculated heat of reaction 12. Bond strengths are from Table VI.
models for complicated heterogeneous catalysts. Ignoring entropic effects, which are expected to be small for these systems, reactions 12 and 13 will be facile only if they are exothermic. The exothermicity requirement leads to restrictions on the bond strengths to oxygen and to methylene. Values for gas-phase ions are listed in Table VI. Note that metathesis with ammonia,a as observed for both GO2+and Reo2+,is facilitated by the relative weakness of the N-H bonds in NH3: O c M 4 + NH3 O-M-NH H20 D(H) = D ( M 4 ) - D(M=NH) - 22 kcal/mol (14) +
+
Thermochemical data and requirements are summarized graphically in Figure 5. Metal-oxo bonds weaken dramatically across a period, whereas metal-methylene bonds weaken only slightly. This is consistent with the ionic character of the oxo bond and the greater electronegativity of the later transition metal^.^^^^ In Figure 5, the dashed horizontal line indicates the energy required to convert methanol to methane and an oxygen atom. Any metal-oxo bond that lies above this line is therefore too strong for the direct conversion of methane to methanol. The triangles indicate the calculated heat for reaction 12. Values above zero (the solid horizontal line) therefore indicate that the doublebond-metathesisreaction is thermodynamicallyunfavorable. To summarize, the later metals favor both reactions 12 and 13 because their bonds to oxygen are weak.s0 (48)Labinger, J. A.; & r a w , J. E. Organometallics 1988, 7, 926-928. (49)Carter, E. A.; Goddard, W. A., 111. J. Phys. Chem. 1988, 92,
2109-21 IS.
(SO) CrO' mcta with ethane to produce ethanol: Kang, H.; Beauchamp,
J. L. J. Am. Chrm. Soc. 1986, 108,7502-7509.
Conclusions Gas-phase studies of the third-row transition-metal cations have revealed reactivity far surpassing that observed in the first two rows. As one progresses from the first to the second to the third row, dehydrogenation becomes increasingly dominant and extensive, until in the third row oligomerization of methane is ob served. Despite the strong spin-orbit coupling, chemical concepts established through studies of the lighter transiton metals remain applicable in the third row. Promotion and exchange energies are still helpful for understanding the pattern of reactivity across the third row. The sizes of the valence orbitals, and also the difference in size between the valence s and d orbitals, are important for transition-metal ions generally. These considerations also appear to be applicable in the presence of simple ligands and for doubly charged ions. Rapid reactions of Ta+ and of W+lead to MC4H8+species, which are probably metallacyclopentanes. Some thermochemical information has been obtained, including D(M+-CH2) = 102 f 9 kcal/mol (M = Re, Hf) and D(M+-CH2) > 111 kcal/mol (M = W, Ir, Pt). Reactions in mixtures of CH4 and O2 provide evidence for carbon-oxygen bond formation. Studies of these or similar systems are expected to lead to gas-phase models for the direct, catalytic conversion of methane into useful products. Acknowledgmenr. This work is supported by the National Science Foundation (Grant CHE 87 11567), the Office of Naval Research (Grant N00014-89-5-3198). the Caltech Consortium in Chemistry and Chemical Engineering (founding members: E. I. duPont de Nemours and Co., Inc.; Eastman Kodak Co.; Minnesota Mining and Manufacturing Co.; Shell Development Co.), and the donors of the Petroleum Research Fund, administered by the American Chemical Society. K.K.I. is grateful to E. H. Fowles for many valuable suggestions and to the Department of Education for fellowship support. Registry No. La+, 14175-57-6;HfC, 20561-33-5;Ta+, 20561-66-4;
W+,16557-44-1;Re+, 20561-58-4;Os+,20561-52-8;Ir+, 54923-08-9; ,'tP 20561-56-2;Au+, 20681-14-5; Cr+, 14067-03-9; Mo+, 16727-12-1; methane. 74-82-8.
