Binuclear and Trinuclear Chromium Carbonyls with Linear Bridging

Mar 17, 2010 - ... of Binuclear Transition Metal Carbonyl Cluster Ions. Guan-jun Wang , Ming-fei Zhou. Chinese Journal of Chemical Physics 2018 31 (1)...
0 downloads 0 Views 3MB Size
Download PDF
4672

J. Phys. Chem. A 2010, 114, 4672–4679

Binuclear and Trinuclear Chromium Carbonyls with Linear Bridging Carbonyl Groups: Isocarbonyl versus Carbonyl Bonding of Carbon Monoxide Ligands Zhong Zhang,† Qian-shu Li,*,†,‡ Yaoming Xie,§ R. Bruce King,*,†,§ and Henry F. Schaefer III§ Center for Computational Quantum Chemistry, South China Normal UniVersity, Guangzhou 510631, People’s Republic of China, Institute of Chemical Physics, Beijing Institute of Technology, Beijing 100081, People’s Republic of China, and Department of Chemistry and Center for Computational Chemistry, UniVersity of Georgia, Athens, Georgia 30602 ReceiVed: NoVember 23, 2009; ReVised Manuscript ReceiVed: January 28, 2010

Isocarbonyl groups bonded to chromium through their oxygen atoms as well as the usual carbonyl groups bonded to carbon atoms have been considered as possible structural features for Cr(CO)6, Cr2(CO)n (n ) 11, 10, 9), and Cr3(CO)16. In this connection, Cr(CO)6 structures with one of the six CO groups bonded to the chromium either solely through its oxygen atom or side-on through both its carbon and oxygen atoms are predicted by density functional theory to lie >30 kcal/mol above the well-known Cr(CO)6 structure, in which all six carbonyl groups are bonded to the chromium atom in the normal manner through their carbon atoms. The binuclear Cr2(CO)11 structure of the type (OC)5Cr-C-O-Cr(CO)5 with a linear bridging carbonyl group bonded to one chromium atom through its carbon atom and to the other chromium atom through its oxygen atom is of lower energy than previously studied Cr2(CO)11 structures and indeed is viable with respect to dissociation into Cr(CO)5 + Cr(CO)6. Similar binuclear structures are found for Cr2(CO)10 and Cr2(CO)9 with linear bridging carbonyl groups. However, the Cr2(CO)9 structure with a linear bridging carbonyl group, no Cr-Cr bond, and 16-electron chromium configurations is found to lie higher by 22 ( 5 kcal/mol than the previously found Cr2(CO)6(µ-CO)3 structure with three bridging carbonyl groups, a formal CrtCr triple bond, and the favorable 18-electron chromium configuration. For the trinuclear Cr3(CO)16, nearly degenerate transand cis-Cr(CO)4[OCCr(CO)5]2 structures are found that are viable with respect to dissociation into 2Cr(CO)5 + Cr(CO)6. 1. Introduction The binuclear carbonyls Co2(CO)8, Fe2(CO)9, and Mn2(CO)10 are well-known and relatively stable isolable transition-metal compounds.1-3 However, the next member of this series, namely, Cr2(CO)11, has never been isolated or even detected spectroscopically in low-temperature matrix studies involving the reaction of chromium atoms with carbon monoxide4 or the photolysis of Cr(CO)6.5 These negative observations were rationalized by a theoretical study nearly a decade ago,6 which provided an explanation of this nonexistence of Cr2(CO)11 by predicting its thermodynamic instability with respect to fragmentation into Cr(CO)6 + Cr(CO)5. The above-noted 2001 theoretical study6 considered only possible Cr2(CO)11 structures having all terminal CO groups and structures in which one to three of the carbonyl groups bridge a Cr-Cr bond. In these structures, the chromium atoms are bonded only to the carbon atoms of the CO groups, whether the CO groups are terminal or bridging. However, a recent theoretical study7 on the binuclear chromium thiocarbonyl analogue Cr2(CS)2(CO)9 predicted low-energy structures containing rare linear Cr-C-O-Cr or bent Cr-C-S-Cr units in which a carbonyl or thiocarbonyl group, respectively, bridges two chromium atoms without a chromium-chromium bond. In such bridges, the CO or CS group is bonded to chromium atoms through both its carbon atom and its chalcogen atom, in contrast * To whom correspondence should be addressed. † South China Normal University. ‡ Beijing Institute of Technology. § University of Georgia.

Figure 1. The CO group bridging a metal-metal bond is the usual type of bridging CO group found in binuclear metal carbonyls. The linear bridging CO (or CS) group with no metal-metal bond is rare in metal carbonyl chemistry but is found in some of the structures predicted for Cr2(CS)2(CO)9 and experimentally in (η5-C5Me5)2 V-O-C-V(CO)5.

to the usual type of bridging CO or CS group, where both metals are bonded only to the carbon atom (Figure 1). Such M-CS-M′ bridging CS groups are found in binuclear metal thiocarbonyl derivatives, including the experimentally known8 [Ph2PCH2CH2PPh2]2(CO)W-CdS-W(CO)5. The latter compound is closely related to Cr2(CO)11 by replacement of one CO group with a CS group, four more CO groups with two chelating ditertiary phosphine ligands, and chromium with tungsten. A similar linear V-C-O-V unit was found by Osborne and co-workers9 in (η5-Me5C5)2V-O-C-V(CO)5. In the latter compound, the bridging CO functions as a normal carbonyl group toward the V(CO)5 unit by bonding to the vanadium through the carbon atom but as an “isocarbonyl” group toward the (η5-Me5C5)2V unit by bonding to the vanadium through the oxygen atom. The prediction of low-energy Cr2(CS)2(CO)9 structures with bridging Cr-C-O-Cr or Cr-C-S-Cr groups and the isola-

10.1021/jp911120y  2010 American Chemical Society Published on Web 03/17/2010

Isocarbonyl versus Carbonyl Bonding of CO Ligands tion of [Ph2PCH2CH2PPh2]2(CO)W-CdS-W(CO)5 as a stable compound raises the question as to whether low-energy homoleptic Cr2(CO)11 structures can be found with similar linear Cr-C-O-Cr structural units. That type of structure was not considered in the earlier study.6 In addition, our previous work on Cr2(CO)10 (ref 10) and Cr2(CO)9 (ref 11) did not consider structures with linear Cr-C-O-Cr units. Accordingly, we have now investigated possible structures containing such linear Cr-C-O-Cr units not only for Cr2(CO)n (n ) 11, 10, 9) but also for the trinuclear derivative Cr3(CO)16, considered as Cr(CO)4[Cr(CO)6]2. In the latter case, we found both cis and trans isomers with essentially the same energies. A comparison is also made between the well-known very stable “carbonyl” structure Cr(CO)6 and the corresponding, but much higher energy, “isocarbonyl” structure Cr(CO)5(OC). 2. Theoretical Methods Electron correlation effects were considered using density functional theory (DFT) methods, which have evolved as a practical and effective computational tool, especially for organometallic compounds.12-26 Two DFT methods were used in this study. The first functional is the popular B3LYP method, which is the hybrid HF/DFT method using a combination of the three-parameter Becke functional (B3) with the LeeYang-Parr (LYP) generalized gradient correlation functional.27,28 The other DFT method used in the present paper is BP86, which combines Becke’s 1988 exchange functional (B) with Perdew’s 1986 gradient corrected correlation functional method (P86).29,30 It has been noted elsewhere that the BP86 method may be more reliable than B3LYP for some properties of the type of organometallic systems considered in this paper.31-33 Our DZP basis sets used for carbon and oxygen add one set of pure spherical harmonic d functions with orbital exponents Rd(C) ) 0.75 and Rd(O) ) 0.85 to the standard HuzinagaDunning contracted DZ sets, designated as (9s5p1d/4s2p1d).34-36 For Cr, in our loosely contracted DZP basis set, the Wachter’s primitive set37 is used, augmented by two sets of p functions and one set of d functions and contracted following Hood et al.,38 and designated (14s11p6d/10s8p3d). For Cr(CO)6, Cr(CO)5, Cr2(CO)11, Cr2(CO)10, Cr2(CO)9, and Cr3(CO)16, there are 229, 199, 428, 398, 368, and 627 contracted Gaussian basis functions, respectively. The geometries of all structures were fully optimized using the B3LYP/DZP and BP86/DZP methods. Vibrational frequencies were determined by evaluating analytically the second derivatives of the energy with respect to the nuclear coordinates. The corresponding infrared intensities were also evaluated analytically. All of the computations were carried out with the Gaussian 03 program,39 exercising the fine grid option (75 radial shells, 302 angular points) for evaluating integrals numerically,40 while the tight (10-8 hartree) designation is the default for the self-consistent field (SCF) convergence. The harmonic vibrational frequencies and their infrared intensities for all of the structures were evaluated using both the B3LYP and BP86 methods. These results were initially used to determine whether a structure is a genuine minimum. The predicted ν(CO) harmonic vibrational frequencies and infrared intensities for the structures are of particular interest since the theoretical results can be compared with the experimentally known compounds. Thus, any future experimental work to detect such species is likely to rely on the relatively strong ν(CO) vibrational frequencies for initial product characterization. These ν(CO) stretching frequencies were obtained with the BP86 method, which has been shown to be more reliable than the B3LYP method for such infrared frequencies.15,41

J. Phys. Chem. A, Vol. 114, No. 13, 2010 4673 Morokuma energy decomposition analyses (EDA)42,43 were carried out for selected BP86 structures at the BP86/DZP level using the program package ADF.44 Thus, the bonding interactions between the C4V fragment Cr(CO)5 and CO, Cr(CO)6, or Cr2(CO)11 were analyzed using the EDA method. In this connection, the bonding dissociation energy is divided into two physically appealing entities

∆E ) ∆Eprep + ∆Eint The preparation energy (∆Eprep) is the energy required to promote the two fragments (A and B) from their isolated equilibrium geometry to the geometry that they have in the compound AB. The interaction energy (∆Eint) is the interaction between the two prepared fragments in the molecule. In this study, ∆Eprep is sufficiently small to be neglected, so that only ∆Eint needs to be considered. This interaction energy can be separated into three major components

∆Eint ) ∆Eels + ∆EPauli + ∆Eorb In this equation, the first term ∆Eels is the electrostatic interaction energy between the fragments, which is calculated using a frozen electron density distribution at the geometry of the complex. The second term ∆EPauli is the repulsive energy caused by Pauli repulsion. The last term ∆Eorb is the stabilization energy from relaxed orbital interaction between the prepared fragments. The ∆Eorb term can be further broken down into orbital contributions from different irreducible representations. This allows the prediction of separate energy contributions from σ and π interactions. 3. Results and Discussion 3.1. Mononuclear Compounds. The global minimum of Cr(CO)6, namely, structure 6a (Figure 2), is found to be a genuine minimum and corresponds to the structure well-known experimentally. The predicted Cr-C distance of 1.923 (B3LYP) or 1.907 Å (BP86) is close to the average Cr-C distance of 1.914 Å found experimentally by neutron diffraction in crystalline Cr(CO)6.45 However, the average C-O distance of 1.140 Å found by neutron diffraction is shorter than our predicted C-O distance of 1.155 (B3LYP) or 1.169 Å (BP86). The dissociation energy (Table 1) of the Cr(CO)6 structure 6a into Cr(CO)5 and CO fragments of 40.4 (B3LYP) or 46.1 kcal/mol (BP86) is consistent with the experimentally determined 37 kcal/ mol.46 Our predicted infrared-active ν(CO) frequency of 1981 cm-1 for 6a is in excellent agreement with the experimentally observed5 value of 1985.4 cm-1 (Table 2). The Cr(CO)6 structure 6b, with one isocarbonyl group bonded to the chromium through oxygen rather than carbon (Figure 2), lies 29.9 (B3LYP) or 34.8 kcal/mol (BP86) above the global minimum 6a, with no imaginary vibrational frequencies. The dissociation energy of 6b into Cr(CO)5 + CO (Table 1) is predicted to be 10.5 (B3LYP) or 11.4 kcal/mol (BP86). This suggests that the Cr(CO)5(OC) structure 6b is viable with respect to such fragmentation but that the Cr-O bonding to the isocarbonyl group is about four times weaker than the Cr-C interaction to a normal carbonyl group. The Cr-O distance in 6b is 2.189 (B3LYP) or 2.136 Å (BP86), and the Cr-C distance trans to the isocarbonyl group shortens to 1.865 (B3LYP) or 1.847 Å (BP86). The Cr(CO)5(OC) isocarbonyl structure 6b was first suggested by Ku¨ndig and Ozin in their matrix isolation studies.4 However,

4674

J. Phys. Chem. A, Vol. 114, No. 13, 2010

Zhang et al.

Figure 2. The three optimized Cr(CO)6 structures.

TABLE 1: Dissociation Energies (in kcal/mol) for the Reported Structures Cr(CO)6 (6a) f Cr(CO)5 + CO (OC)5Cr(OC) (6b) f Cr(CO)5 + CO Cr2(CO)11 (11a) f Cr2(CO)10 (10a) + CO Cr2(CO)10 (10a) f Cr2(CO)9 (9a) + CO Cr2(CO)11 (11a) f Cr(CO)6 + Cr(CO)5 Cr2(CO)10 (10a) f Cr(CO)6 + Cr(CO)4 Cr2(CO)10 (10a) f Cr(CO)5 + Cr(CO)5 Cr2(CO)9 (9a) f Cr(CO)6 + Cr(CO)3 Cr2(CO)9 (9a) f Cr(CO)4 + Cr(CO)5 Cr3(CO)16 (16a) f Cr(CO)6 + 2Cr(CO)5

B3LYP

BSSE

BP86

BSSE

40.4 10.5 36.8 21.3 9.9 12.8 13.4 13.0 15.0 18.8

4.6 4.8

46.1 11.4 41.5 16.9 10.5 13.7 15.1 13.5 15.7 20.3

4.6 4.8

3.1 3.1

3.1 3.1

TABLE 2: ν(CO) Stretching Frequencies Predicted for the Cr(CO)n (n ) 6, 5) Derivativesa ν(CO) 6a (Oh) exp. (ref 5) 6b (C4V) 6c (Cs) 5a (C4V) exp. (ref 5) 5b-t (D3h) exp. (ref 4) a

2086(a1g, 0), 2001(eg, 0), 2001(eg, 0), 1981(t1u, 1603), 1981(t1u, 1603), 1981(t1u, 1603) 1985(t1u) 2063(a1, 101), 2010(a1, 8), 1987(b2, 0), 1963(e, 1698), 1963(e, 1698), 1957(a1, 727) 2067(a′, 91), 1997(a′, 1), 1980(a′, 1558), 1968(a′′, 1648), 1964(a′, 649), 1887(a′, 199) 2061(a1, 6), 1983(b2, 0), 1958(e, 1771), 1958(e, 1771), 1949(a1, 727) 2088(a1), 1961(e), 1932(a1) 2046(a1′, 0), 1969(e, 1050), 1969(e, 1050), 1964(a1′, 0), 1952(a2′′, 1760) 1964(a2′′), 1937(e)

Infrared intensities are given in parentheses in km/mol.

Turner and co-workers5 concluded that bands attributable to the isocarbonyl group are absent in these matrixes and that, in addition, it is impossible to reconcile the sensitivity of the Cr(CO)5 visible absorption band to the matrix with such a structure. The predicted ν(CO) frequencies (Table 2) suggest that the isocarbonyl structure 6b nevertheless might exist in such matrixes since its predicted that the ν(CO) spectrum might be masked by the ν(CO) frequencies of the two isomers of the major photoproduct Cr(CO)5. In particular, the strong predicted ν(CO) frequencies of 1963 and 1957 cm-1 of the isocarbonyl Cr(CO)5(OC) (6b) could be hidden by the strong 1958 cm-1 ν(CO) frequency of the square pyramidal Cr(CO)5 isomer 5a and the strong 1969 cm-1 ν(CO) frequency of the trigonal bipyramidal Cr(CO)5 isomer 5b-t (Table 2). The Cr(CO)6 structure 6c with one side-on-bonded η2-CO ligand (Figure 2) is predicted to lie 33.8 (B3LYP) or 36.8 kcal/ mol (BP86) above 6a with no imaginary vibrational frequencies. The calculated energy for dissociation of 6c into Cr(CO)5 + CO is 6.3 (B3LYP) or 9.4 kcal/mol (BP86), suggesting modest thermodynamic viability toward dissociation. Ku¨ndig and Ozin4 detected a rather low ν(CO) frequency of 1892 cm-1 in their

Figure 3. The two optimized structures for Cr(CO)5.

matrixes, which is close to the predicted 1887 cm-1 ν(CO) frequency of 6c (Table 1). Such a low ν(CO) frequency must either arise from a side-on-bonded η2-CO ligand in a mononuclear chromium carbonyl derivative (such as 6c) or a bridging carbonyl group in a chromium carbonyl derivative with two or more chromium atoms. The coordinately unsaturated Cr(CO)5 has been generated in low-temperature matrices either by reaction of chromium atoms with carbon monoxide4 or by photolysis of Cr(CO)6.5 The lowest-energy structure 5a for Cr(CO)5 (Figure 3) is a genuine minimum, derived from the hexacoordinate global minimum Cr(CO)6 structure 6a by loss of a CO group, and thus has square pyramidal chromium coordination. This Cr(CO)5 structure has been observed experimentally in low-temperature matrices.4,5 In addition, a triplet-state Cr(CO)5 structure 5b-t (Figure 3) with trigonal bipyramidal chromium coordination is found at 14.2 (B3LYP) or 19.0 kcal/mol (BP86) above the singlet global minimum 5a. The predicted infrared-active ν(CO) harmonic vibrational frequencies for the C4V square pyramidal Cr(CO)5 structure 5a (Figure 2 and Table 2) of 2061, 1958, and 1949 cm-1 are close to the experimental5 ν(CO) frequencies of 2088, 1961, and 1932 cm-1. Furthermore, the predicted infrared-active ν(CO) frequencies for the D3h trigonal bipyramidal Cr(CO)5 structure 5b-t of 1969 and 1952 cm-1 are close to the experimental values of 1964 and 1937 cm-1 assigned by Ku¨ndig and Ozin4 to trigonal bipyramidal Cr(CO)5. However, this latter assignment was questioned in a subsequent reinterpretation of Ku¨ndig and Ozin’s data4 by Black and Braterman.47 3.2. Binuclear Compounds Cr2(CO)n (n ) 11, 10, 9). Our earlier theoretical work6 reported a doubly semibridged structure 11 (Figure 4) for the global minimum of Cr2(CO)11 using the BP86 method. This structure was reported to be slightly thermodynamically unfavored by 1.7 kcal/mol with respect to dissociation into Cr(CO)6 + Cr(CO)5.

Isocarbonyl versus Carbonyl Bonding of CO Ligands

Figure 4. The optimized Cr2(CO)11 structures.

TABLE 3: ν(CO) Stretching Frequencies Predicted for the Binuclear Cr2(CO)n (n ) 11, 10, 9) Derivativesa ν(CO) Cr2(CO)11 11a (C4V) 2085(a1, 49), 2053(a1, 229), 2008(b2, 0), 2004(a1, 423), 1989(e, 1888), 1989(e, 1888), 1983(b1, 0), 1956(e, 1060), 1956(e, 1060), 1955(a1, 145), 1948(a1, 2370) Cr2(CO)10 10a (Cs) 2083(a′, 154), 2018(a′, 313), 2011(a′′, 14), 2006(a′, 803), 1992(a′′, 1642), 1992(a′, 1515), 1936(a′, 349), 1931(a′′, 1368), 1919(a′, 823), 1915(a′, 1644) 10b (C4V) 2068(a1, 0), 2053(a1, 98), 1994(b1, 0), 1979(b2, 0), 1971(e, 2455), 1971(e, 2455), 1951(e, 634), 1951(e, 634), 1946(a1, 431), 1927(a1, 1778) 10c (Cs) 2160(a′, 158), 2100(a′, 1086), 2088(a′, 474), 2084(a′, 711), 2079(a′′, 2229), 2055(a′, 1345), 2037(a′, 974), 2020(a′′, 781), 2018(a′, 169), 1904(a′, 410) 10d (Cs) 2063(a′, 12), 2034(a′, 1151), 1985(a′′, 16), 1985(a′, 576), 1969(a′′, 2954), 1963(a′, 2166), 1956(a′′, 106), 1954(a′, 127), 1953(a′, 843), 1910(a′, 886) Cr2(CO)9 9a (Cs) 2060(a′, 219), 2004(a′, 1324), 1998(a′′, 15), 1974(a′, 1785) 1973(a′′, 1931), 1928(a′, 822), 1926(a′′, 1257), 1914(a′, 838), 1867(a′, 225) 9b (C4V) 2050(a′, 66), 2009(a′, 1494), 1975(a′, 945), 1966(a′′, 26) 1960(a′′, 2314), 1951(a′, 1192), 1932(a′, 1293), 1931(a′′, 987), 1895(a′, 150) 9c (Cs) 2063(a1, 12), 2022(a1, 103), 1990(b2, 0), 1964(e, 2090) 1964(e, 2090), 1957(b1, 0), 1937(a1, 987), 1919(e, 1302), 1919(e, 1302) a Infrared intensities are given in parentheses in km/mol; the ν(CO) frequencies of the Cr-C-O-Cr units are in bold.

Structures with linear Cr-C-O-Cr units were not considered in our earlier work.6 Optimizing a structure of this type leads to the C4V structure 11a (Figure 4) as a genuine minimum with all real vibrational frequencies. The total energy of 11a is predicted to be 12.2 kcal/mol (BP86) below structure 11. Furthermore, the predicted energy of 9.9 (B3LYP) or 10.5 kcal/ mol (BP86) for the dissociation of 11a into Cr(CO)6 (6a) plus Cr(CO)5 (5a) fragments (Table 1) indicates the thermodynamic stability of 11a. Structure 11a has a linear Cr-C-O-Cr unit with a predicted Cr-O(C) distance of 2.201 (B3LYP) or 2.153 Å (BP86), which is quite similar to that of 6b (Figure 2). The Cr2(CO)11 structure 11a is predicted to have two weak high ν(CO) frequencies of 2085 and 2053 cm-1, two more intense harmonic vibrational frequencies of 2004 and 1989 cm-1, as well as three low ν(CO) frequencies of 1956, 1955, and 1948 cm-1 (Table 3). The strong 1956 cm-1 ν(CO) frequency corresponds to the asymmetric vibration of the four equatorial carbonyls of the Cr(CO)5(OC) unit of 11a. The 1955 (weak) and 1948 cm-1 (strong) ν(CO) frequencies correspond to the asymmetric vibration of the three axial carbonyls (Table 3). A total of four new structures were obtained for Cr2(CO)10 (Figure 5 and Table 4). The previously optimized10 lowestenergy Cr2(CO)10 structure 10 is included for comparison.

J. Phys. Chem. A, Vol. 114, No. 13, 2010 4675 The new Cr2(CO)10 structure 10a with a linear four-electrondonor bridging CO group (Figure 5 and Table 4) is found to have a tiny imaginary frequency of 4i (B3LYP) or 6i cm-1 (BP86). However, when an ultrafine integration grid (99, 590) is used for the B3LYP optimization of 10a, this imaginary frequency decreases to 2i cm-1, indicating that this structure is either a genuine minimum or very close to a genuine minimum. The Cr2(CO)10 structure 10a is derived from the Cr2(CO)11 structure 11a by loss of an equatorial CO group from the Cr(CO)5 fragment. The predicted dissociation energy of 10a with respect to dissociation into Cr(CO)6 + Cr(CO)4 (Table 1) is 12.8 (B3LYP) or 13.7 kcal/mol (BP86), which is slightly larger than that of 11a. The predicted Cr-O(C) distance in 10a of 2.125 (B3LYP) or 2.080 Å (BP86) is somewhat shorter than that of 2.201 (B3LYP) or 2.153 Å (BP86) in 11a, suggesting that the strength of the Cr-O bond in 10a linking the two halves is greater upon loss of a CO group from 11a. This new Cr2(CO)10 structure 10a is predicted to lie 4.9 kcal/mol below the previously reported10 global minimum 10 by B3LYP but 1.4 kcal/mol above 10 by BP86. The next Cr2(CO)10 structure, namely, the C4V structure 10b (Figure 5 and Table 4), is derived from 11a (Figure 4) by loss of an axial CO group from the Cr(CO)5 fragment. Structure 10b lies 2.9 kcal/mol below 10 by B3LYP but 4.7 kcal/mol above 10 by BP86. The last Cr2(CO)10 structure with a linear Cr-C-O-Cr unit, namely, 10d, is derived from 11a by loss of an equatorial CO group from the Cr(CO)6 fragment. Structure 10d lies 1.8 kcal/mol below 10 by B3LYP but 5.0 kcal/mol above 10 by BP86. These three Cr2(CO)10 structures 10a, 10b, and 10d with linear Cr-C-O-Cr units and without any direct Cr-Cr interactions (Figure 5 and Table 4) have an 18-electron configuration for one chromium atom but only a 16-electron configuration for the other chromium atom. The linear bridging CO groups in these structures are predicted to exhibit significantly lower ν(CO) frequencies (1920 ( 10 cm-1) than the terminal carbonyl groups (Table 3). A singly bridged Cs structure 10c is also found for Cr2(CO)10 at 2.1 (B3LYP) or 1.2 kcal/mol below 10 (Figure 5 and Table 3). The relatively short Cr-O distance of ∼2.37 Å in 10c indicates a four-electron-donor η2-µ-CO group. The predicted Cr-Cr distance of ∼3.1 Å in 10c suggests a formal single bond, thereby giving both chromium atoms the favored 18-electron configuration in a Cr2(CO)10 structure with one 4-electron-donor carbonyl group. This η2-µ-CO group in 10c is predicted to exhibit a ν(CO) frequency at 1904 cm-1. The lowest-energy structure for Cr2(CO)9 with a linear Cr-C-O-Cr unit, namely, 9a (Figure 6 and Table 5), is predicted to have a tiny imaginary frequency of 6i (B3LYP) or 7i cm-1 (BP86). However, this imaginary frequency drops to 1i cm-1 when 9a is reoptimized with the B3LYP method using an ultrafine grid (99, 590). The Cr2(CO)9 structure 9a is generated by joining C2V Cr(CO)4 and C4V Cr(CO)5 fragments with a linear CO bridge. Structure 9a lies 16.8 or 27.2 kcal/ mol above the triply bridged global minimum of Cr2(CO)9 found in the previous work.11 However, the energy for dissociation of 9a into Cr(CO)4 + Cr(CO)5 fragments (Table 1) of 15.0 (BP86) or 15.7 kcal/mol (BP86) is somewhat larger than that of 10a, indicating that this structure is thermodynamically stable with a stronger Cr-O interaction in 9a than in 10a. The remaining two genuine minimum Cr2(CO)9 structures 9b and 9c (Figure 6 and Table 5) can be constructed by joining D4h Cr(CO)4 and C4V Cr(CO)5 fragments though linear Cr-C-O-Cr bridges in different ways. These structures are

4676

J. Phys. Chem. A, Vol. 114, No. 13, 2010

Zhang et al.

Figure 5. Four new optimized Cr2(CO)10 structures.

TABLE 4: Total Energies (E, in Hartree), Relative Energies (∆E, in kcal/mol), and Number of Imaginary Frequencies (Nimg) for the Optimized Cr2(CO)10 Structuresa 10 (C2h)

10a (Cs)

10b (C4V)

10c (Cs)

911

9a (Cs)

9b (C4V)

9c (Cs)

3109.28953 0.0 0 3109.71158 0.0 21i

3109.26272 16.8 6i 3109.66823 27.2 7i

3109.24080 30.6 0 3109.63864 45.8 0

3109.24040 30.8 0 3109.64084 44.4 0

10d (Cs)

B3LYP –E 3222.64446 3222.65218 3222.64907 3222.64775 3222.64734 ∆E 0.0 –4.9 –2.9 –2.1 –1.8 Nimg 0 4i 0 0 0 B3LYP –E 3223.06812 3223.06582 3223.06053 3223.06997 3223.06004 ∆E 0.0 1.4 4.7 –1.2 5.0 Nimg 0 6i 0 0 0 a

TABLE 5: Total Energies (E, in Hartree), Relative Energies (∆E, in kcal/mol), and Number of Imaginary Frequencies (Nimg) for the Optimized Cr2(CO)9 Structures

The conventional structure 10 is described in ref 10.

predicted to lie above the global minimum of Cr2(CO)9 by more than 30.0 kcal/mol with both methods. 3.3. Trinuclear Compounds. The two structures found for the trinuclear complex Cr3(CO)16 (Figure 7) can be regarded as derived from Cr(CO)6 by replacement of two carbonyl groups with Cr(CO)5(OC) (6b) “ligands” bonded to the central chro-

B3LYP

B3LYP

-E ∆E Nimg -E ∆E Nimg

mium atom through the carbon of their isocarbonyl groups. The D4h structure 16a is the trans isomer of this type, whereas the Cs structure 16b is the cis structure of this type. All three chromium atoms in both 16a and 16b have the favored 18electron configuration. The D4h trans-Cr3(CO)16 structure 16a (Figure 7) is found to have a small imaginary vibrational frequency of 5i cm-1 with both methods. This structure is a genuine local minimum since

Figure 6. Three new optimized Cr2(CO)9 structures, all with linear Cr-C-O bridging units.

Isocarbonyl versus Carbonyl Bonding of CO Ligands

J. Phys. Chem. A, Vol. 114, No. 13, 2010 4677

Figure 7. Two optimized Cr3(CO)16 structures.

Figure 8. The forward σ bonding and π back-bonding of a CO group to a Cr(CO)5 fragment to form Cr(CO)6.

the small imaginary frequency disappears upon reoptimization of the B3LYP geometry using the ultrafine (99, 590) numerical integration grid. Structure 16a is predicted to be thermodynamically stable with respect to dissociation into a Cr(CO)6 and two Cr(CO)5 fragments since the calculated dissociation energy of 16a is 18.8 (B3LYP) or 20.3 kcal/mol (BP86), nearly twice that of 11a. The Cs cis-Cr3(CO)16 structure 16b (Figure 7), like 16a, also has a tiny imaginary vibrational frequency of 1i (B3LYP) or 5i cm-1 (BP86). Structure 16b is predicted to lie only 0.1 (B3LYP) or 0.2 kcal/mol above the corresponding trans structure 16a, indicating that these two Cr3(CO)16 stereoisomers are essentially degenerate. The highly symmetric trans-Cr3(CO)16 16a has only five distinct infrared-active ν(CO) frequencies (Table 5), among which the two lowest frequencies (1956 and 1946 cm-1) can be assigned to the vibration of the four axial carbonyls. The pattern of vibrational frequencies for the much less symmetrical cis-Cr3(CO)16 structure 16b appears to be quite complicated since all 16 ν(CO) are infrared-active. The Cr-O(C) distances in structures 11a (Figure 4), 16a, and 16b (Figure 7) are predicted to be ∼2.20 (B3LYP) or ∼2.15 Å (BP86), which are slightly longer than that in 6b. The C-O distances in the linear bridging carbonyl groups of structures 11a, 16a, and 16b are predicted to be 1.161 (B3LYP) or 1.175 Å (BP86), somewhat longer than that of that of the terminal isocarbonyl group in 6b, namely, 1.146 (B3LYP) or 1.160 Å (BP86). This can arise from the dπ f pπ* back bonding of the second chromium atom to the antibonding

orbitals of the linear bridging carbonyl group in the binuclear and trinuclear derivatives. The geometrical parameters of the various bond types in the Cr2(CO)11 structure 11a and the trinuclear Cr3(CO)16 structures 16a and 16b are nearly the same. 3.4. Chemical Bonding. The familiar transition-metal carbonyl interaction is often described by the popular DewarChatt-Duncanson (DCD) model48,49 of M r CO σ donation and M f CO π back-donation (Figure 8). Similar orbital interactions are possible for isocarbonyl ligands in which the metal is bonded to the oxygen atom of the CO group rather than the carbon atom. This is suggested by the MOs of the chromium isocarbonyl derivatives discussed in this paper. Thus, for Cr(CO)5(OC) (6b in Figure 2), the σ bond is formed by overlap of the a1 orbital of Cr(CO)5 with the lone pair orbital of the oxygen atom in the isocarbonyl ligand. The degenerate pair of π bonds are formed though overlap of the perpendicular pair of e1 orbitals of the Cr(CO)5 fragment with the π* antibonding orbitals of the CO ligands. Similar orbital interactions (Figure 9) are found for (OC)5CrCOCr(CO)5 (11a in Figure 4) and (OC)4Cr[OCCr(CO)5]2 (16a in Figure 7). Table 7 shows that the calculated interaction energies (∆Eint) of 6b, 11a, and 16a are quite close to the predicted dissociation energies with respect to 5a and other fragments. The interaction energies of the chromium isocarbonyl (CrOC) units in 6b, 11a, and 16a fall in the range of 11.0-12.5 kcal/mol. This corresponds to approximately one-fourth of those for the chromium carbonyl (Cr-CO) interactions in 6a. In addition, the longer Cr-O distances to the isocarbonyl groups in 6b, 11a, and 16a

4678

J. Phys. Chem. A, Vol. 114, No. 13, 2010

Zhang et al.

Figure 9. The forward σ bonding and π back-bonding of a Cr(CO)6 group to a Cr(CO)5 fragment to form the Cr2(CO)11 structure with a linear Cr-C-O-Cr unit.

TABLE 6: ν(CO) Stretching Frequencies Predicted for the Two Cr3(CO)16 Structuresa ν(CO) 16a (D4h)

16b (Cs)

2083(a1g, 0), 2053(a1g, 0), 2053(a2u, 580), 2013(b1g, 0), 1996(eu, 2070), 1996(eu, 2070), 1984(b2g, 0), 1984(b1u, 0), 1974(a1g, 0), 1958(eu, 2180), 1958(eu, 2180), 1958(eg, 0), 1958(eg, 0), 1956(a2u, 157), 1955(a1g, 0), 1946(a2u, 4204) 2082(a′, 93), 2053(a′, 264), 2053(a′, 324), 2011(a′, 304), 2008(a′, 665), 1996(a′′, 2102), 1984(a′′, 3), 1984(a′, 26), 1968(a′, 2055), 1959(a′′, 2028), 1959(a′, 1004), 1958(a′, 79), 1956(a′′, 2), 1954(a′, 1307), 1953(a′, 2415), 1947(a′, 310)

a Infrared intensities are given in parentheses in km/mol; bridging ν(CO) frequencies are in bold type.

TABLE 7: Energy Decomposition of 6a, 6b, 11a, 10a, 9a, and 16a (units in kcal/mol) Predicted by BP86 6a (Oh) 6b (C4V) 11a (C4V) 10a (Cs) 9a (Cs) 16a (D4h) BDE ∆Eint ∆EPaul ∆Eels ∆Ester ∆Eorb a1 a2 b1 b2 e1

–46.1 –45.9 108.6 –76.2 32.4 –75.6 –36.8 0.0 –0.08 –0.07 –38.6

–11.4 –11.0 28.3 –17.6 10.7 –18.7 –9.6 0.0 –0.01 –0.01 –9.0

–10.5 –12.5 26.4 –17.1 9.3 –17.9 –9.6 0.0 –0.02 –0.03 –8.2

–12.8 –9.9 25.3 –18.1 7.3 –17.2

–15.0 –11.0 27.3 –21.2 6.1 –17.2

–15.8

–15.8

–5.7

–5.1

–10.2 –11.8 26.8 –16.1 10.6 –18.4 –9.6 0.0 –0.02 –0.03 –8.8

(∼2.2 Å) than the Cr-C distances to the carbonyl groups in 6b (∼1.9 Å) decrease Pauli repulsion energies, electrostatic attraction energies, and total orbital interaction energies to ∼27, ∼17, and ∼18 kcal/mol, respectively. The stabilization energies for σ donation involving the a1 MO for the isocarbonyl group in 6b, 11a, and 16a are predicted to be ∼9.6 kcal/mol (Table 7), which also is nearly one-fourth of that of 6a. Furthermore, the stabilization energies for π back-donation in the Cr-O bonds of the isocarbonyl groups involving the e1 MO are found to be ∼8.5 kcal/mol, somewhat smaller than those for σ donation. Thus, in terms of molecular orbitals (Figures 8 and 9), the Cr-OC interactions in the isocarbonyl derivatives are similar to the Cr-CO interactions in the carbonyl groups. However, the energies of the Cr-O bonds in the isocarbonyls are roughly 25% of those for the Cr-C bonds involving normal carbonyl groups. 3.5. Energy Surfaces. Is it possible to form homoleptic mononuclear or binuclear chromium carbonyl compounds from experimentally known C4V Cr(CO)5 and CO or Oh Cr(CO)6 fragments? These isomers are predicted to be thermodynamically viable. Reoptimization of the isocarbonyl structures 6b and 11a obtained by the B3LYP method at different frozen Cr-O distances using the B3LYP method (Figure 10) suggests

Figure 10. The structures of the isocarbonyls Cr(CO)5(OC) (6b) and (OC)5Cr(OC)Cr(CO)5 (11a) as optimized at different frozen Cr-O distances.

considerable similarity of the energy surfaces of 6b and 11a. The postulated reaction paths are plausible since these reactions are straightforward without any transition states. 4. Summary The isocarbonyl structure Cr(CO)5(OC) (6b) and the sideon-bonded carbonyl structure Cr(CO)5(η2-CO) (6c) (Figure 2) are both predicted to lie >30 kcal/mol above the isomeric wellknown Cr(CO)6 structure 6a, in which all six carbonyl groups are bonded through their carbon atoms in the usual manner. Because of the high energies of the atypical Cr(CO)6 structures 6b and 6c, it is unlikely that either of these two isomers is present in the mixtures obtained by Ku¨ndig and Ozin4 from reactions of chromium atoms with carbon monoxide in lowtemperature matrices. Other interpretations are more likely for their observed ν(CO) spectra. In view of the improvements in experimental matrix isolation methods since Ku¨ndig and Ozin’s work4 more than 35 years ago, reinvestigation of this system is clearly needed. The coordinately saturated binuclear Cr2(CO)11 structure 11a, containing a linear bridging Cr-C-O-Cr unit and no Cr-Cr bond (Figure 4), is predicted to be viable with respect to dissociation into Cr(CO)6 + Cr(CO)5, in contrast with the Cr2(CO)11 structures found in the previous work.6 The coordinately unsaturated binuclear Cr2(CO)n (n ) 10, 9) structures 10a, 10b, 10d, 9a, 9b, and 9c (Figures 5 and 6) with similar linear Cr-C-O-Cr units are also predicted to be viable. Furthermore, the predicted dissociation energies for coordinately unsaturated binuclear derivatives suggest that the binding energies increase with increasing unsaturation. This is consistent with previously observed results for other types of binuclear Cr2(CO)n (n ) 11, 10, 9) structures, as well as the decreasing

Isocarbonyl versus Carbonyl Bonding of CO Ligands Cr-O distances in these structures upon removal of CO groups. However, for Cr2(CO)9, the previously found11 triply bridged Cr2(CO)6(µ-CO)3 structure 9 (Figure 6) with a formal CrtCr triple bond lies 22 ( 5 kcal/mol below the lowest-energy structure with a Cr-C-O-Cr bridge and no Cr-Cr bond, that is, structure 9a. This may relate to the favorable 18-electron configurations of both chromium atoms in the triply bridged structure 9 with a formal CrtCr triple bond compared to the 16-electron configurations of both chromium atoms in the linearly bridged structure 9a with no Cr-Cr bonding. The trinuclear Cr3(CO)16 structures 16a and 16b (Figure 7) are also favored with respect to dissociation into 2Cr(CO)5 + Cr(CO)6. These structures may be derived from Cr(CO)6 by replacing two carbonyl groups in trans positions for the D4h structure 16a or those in cis positions for the Cs structure 16b by Cr(CO)5(OC) (6b) ligands. These Cr(CO)5(OC) ligands bond to the central chromium atom through the carbon atom of their isocarbonyl group. This leads to trinuclear structures containing two Cr-C-O-Cr units. There is essentially no difference in the predicted relative energies of the trans- and cis-Cr(CO)4[COCr(CO)5]2 stereoisomers. Acknowledgment. We are indebted to the 111 Project (B07012) and the National Natural Science Foundation (20873045 and 20973066) of China as well as the U.S. National Science Foundation (Grants CHE-0749868 and CHE-0716718) for support of this research. We also appreciate the Hongcam Software Company for supporting a trial version ADF2008.1. Supporting Information Available: Tables S1-S3: The theoretical harmonic vibrational frequencies for Cr(CO)m (m ) 6, 5; five structures), Cr2(CO)n (n ) 11, 10, 9; eight structures), and Cr3(CO)16 (two structures) using the BP86 method. Tables S4-S18: The theoretical Cartesian coordinates for Cr(CO)m (m ) 6, 5; five structures), Cr2(CO)n (n ) 11, 10, 9; seven structures), and Cr3(CO)16 (two structures) using the B3LYP method; complete Gaussian 03 reference (ref 39). This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Mond, L.; Hirtz, H.; Cowap, M. D. J. Chem. Soc. 1910, 57, 798. (2) Dewar, J.; Jones, H. O. Proc. R. Soc. London, Ser. A 1907, 79, 66. (3) Brimm, E. O.; Lynch, M. A.; Sesny, W. J. J. Am. Chem. Soc. 1954, 76, 3831. (4) Ku¨ndig, E. P.; Ozin, G. A. J. Am. Chem. Soc. 1974, 96, 3820. (5) Perutz, R. N.; Turner, J. J. Inorg. Chem. 1971, 14, 262. (6) Richardson, N. A.; Xie, Y.; King, R. B.; Schaefer, H. F. J. Phys. Chem. A 2001, 105, 11134. (7) Zhang, Z.; Li, Q.-s.; Xie, Y.; King, R. B.; Schaefer, H. F. J. Phys. Chem. A 2010, 114, 486. (8) Dombek, B. D.; Angelici, R. J. J. Am. Chem. Soc. 1974, 96, 7568.

J. Phys. Chem. A, Vol. 114, No. 13, 2010 4679 (9) Osborne, J. H.; Rheingold, A. L.; Trogler, W. C. J. Am. Chem. Soc. 1985, 107, 6292. (10) Li, S.; Richardson, N. A.; Xie, Y. M.; King, R. B.; Schaefer, H. F. Faraday Discuss. 2003, 124, 315. (11) Li, S.; Richardson, N. A.; King, R. B.; Schaefer, H. F. J. Phys. Chem. A 2003, 107, 10118. (12) Ehlers, A. W.; Frenking, G. J. Am. Chem. Soc. 1994, 116, 1514. (13) Delley, B.; Wrinn, M.; Lu¨thi, H. P. J. Chem. Phys. 1994, 100, 5785. (14) Li, J.; Schreckenbach, G.; Ziegler, T. J. Am. Chem. Soc. 1995, 117, 486. (15) Jonas, V.; Thiel, W. J. Chem. Phys. 1995, 102, 8474. (16) Barckholtz, T. A.; Bursten, B. E. J. Am. Chem. Soc. 1998, 120, 1926. (17) Niu, S.; Hall, M. B. Chem. ReV. 2000, 100, 353. (18) Macchi, P.; Sironi, A. Coord. Chem. ReV. 2003, 238, 383. (19) Carreon, J.-L.; Harvey, J. N. Phys. Chem. Chem. Phys. 2006, 8, 93. (20) Bu¨hl, M.; Kabrede, H. J. Chem. Theory Comput. 2006, 2, 1282. (21) Ziegler, T.; Autschbach, J. Chem. ReV. 2005, 105, 2695. (22) Waller, M. P.; Bu¨hl, M.; Geethanakshmi, K. R.; Wang, D.; Thiel, W. Chem.sEur. J. 2007, 3, 4723. (23) Hayes, P. G.; Beddie, C.; Hall, M. B.; Waterman, R.; Tilley, T. D. J. Am. Chem. Soc. 2006, 128, 428. (24) Bu¨hl, M.; Reimann, C.; Pantazis, D. A.; Bredow, T.; Neese, F. J. Chem. Theory Comput. 2008, 4, 1449. (25) Besora, M.; Carreon-Macedo, J.-L.; Cowan, J.; George, M. W.; Harvey, J. N.; Portius, P.; Ronayne, K. L.; Sun, X.-Z.; Towrie, M. J. Am. Chem. Soc. 2009, 131, 3583. (26) Ye, S.; Tuttle, T.; Bill, E.; Simkhorich, L.; Gross, Z.; Thiel, W.; Neese, F. Chem.sEur. J. 2008, 14, 10839. (27) Becke, A. D. J. Chem. Phys. 1993, 98, 5648. (28) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785. (29) Becke, A. D. Phys. ReV. A 1988, 38, 3098. (30) Perdew, J. P. Phys. ReV. B 1986, 33, 8822. (31) See especially Furche, F.; Perdew, J. P. J. Chem. Phys. 2006, 124, 044103. (32) Wang, H. Y.; Xie, Y.; King, R. B.; Schaefer, H. F. J. Am. Chem. Soc. 2005, 127, 11646. (33) Wang, H. Y.; Xie, Y.; King, R. B.; Schaefer, H. F. J. Am. Chem. Soc. 2006, 128, 11376. (34) Dunning, T. H. J. Chem. Phys. 1970, 53, 2823. (35) Dunning, T. H.; Hay, P. J. Methods of Electronic Structure Theory; Schaefer, H. F., Ed.; Plenum: New York, 1977; pp 1-27. (36) Huzinaga, S. J. Chem. Phys. 1965, 42, 1293. (37) Wachters, A. J. H. J. Chem. Phys. 1970, 52, 1033. (38) Hood, D. M.; Pitzer, R. M.; Schaefer, H. F. J. Chem. Phys. 1979, 71, 705. (39) Frisch, M. J.; et al. Gaussian 03, revision C 02; Gaussian, Inc.: Wallingford CT, 2004; (see Supporting Information for details). (40) Papas, B. N.; Schaefer, H. F. J. Mol. Struct. 2006, 768, 175. (41) Silaghi-Dumitrescu, I.; Bitterwolf, T. E.; King, R. B. J. Am. Chem. Soc. 2006, 128, 5342. (42) Morokuma, K. J. Chem. Phys. 1971, 55, 1236. (43) Kitaura, K.; Morokuma, K. Int. J. Quantum Chem. 1976, 10, 325. (44) Fonseca, G. C.; Snijders, J. G.; te Velde, G.; Baerends, E. J. Theor. Chem. Acc. 1998, 99, 391. (45) Rees, B.; Mitschler, A. J. Am. Chem. Soc. 1976, 98, 7918. (46) Lewis, K. E.; Golden, D. M.; Smith, G. P. J. Am. Chem. Soc. 1984, 106, 3905. (47) Black, J. D.; Braterman, P. S. J. Am. Chem. Soc. 1975, 97, 1975. (48) Dewar, M. J. S. Bull. Soc. Chim. Fr. 1951, 18, C79. (49) Chatt, J.; Duncanson, L. A. J. Chem. Soc. 1953, 2929.

JP911120Y