12743
J. Phys. Chem. 1995,99, 12743-12750
Theoretical Study of the Electronic Structure and of the Mercury-Carbon Bonding of Methylmercury(I1) Compounds Vincenzo Barone? Alessandro Bencini,*9$Federico Totti: and Myriam G. Uytterhoeved Dipartimento di Chimica, Universitii Federico II, via Meuocannone 4, 80134 Naples, Italy, and Dipartimento di Chimica, Universitii Degli Studi di Firenze, via Maragliano 75, 50144 Florence, Italy Received: March 2, 1995; In Final Form: May 23, 1 9 9 9
An ab-initio study is presented for the whole series of mercury dihalides and for representative organomercury compounds including the moiety MeHg+ (Me = CH3): MeHgX (X = C1, Br, I), Me2Hg, [MeHg(PH3)]+, and [MeHg(PH3)3]+. The last compound is used to model the [MeHg(np3)]+ (np3 = N(CH2CH2Ph2)3) complex. A basis set of moderate size has been selected, which enables the accurate reproduction of experimental observables like geometries, vibrational frequencies, and thermochemical parameters, while still allowing the study of large molecules. The resistance of the HgC bond to acidic attack has been found to decrease in the order MeHgCl > MeHgBr > MeHgI on the one hand and [MeHg(PH3)]+ > [MeHg(PH3)3]+ on the other hand. In agreement with the experimental observations on [MeHg(np3)]+ the easiest HgC bond cleavage occurs in the [MeHg(PH3)3]+ compound.
I. Introduction The easy methylation of Hg and the stability of the HgMe+ (Me = CH3) moiety in a large variety of chemical environments are well-known. It seems that only upon methylation mercury achieves the ability to link to organic matter of various kinds and tends to accumulate in living organisms at levels of toxicity.' Moreover, cleavage of the HgC bond occurs almost exclusively in rather severe conditions. For instance the linear compounds MeHgX (X = C1, Br, I) need strong mineral acids and high temperatures in order to give methane and the corresponding mercury dihalide.2 Recently, novel compounds of the type [RHg(np3)]+ (R = CH3, C2H5, C6H5 and np3 = N(CH2CHzPh2)3) were de~cribed,~ which contain an Hg atom linked to the R group and to three P atoms of the np3 tripod ligand in a pseudotetrahedral coordination. In contrast to the linear RHgX compounds, the HgC bond in these systems can be cleaved by the action of weak acids, for example acetic acid. The HgC cleavage might be promoted by the strong donating effect of the three phosphorous atoms of the np3 tripod, which increases the negative charge on the carbon atom, thus favoring an electrophilic attack. Altematively, the pseudotetrahedral coordination, rarely observed for Hg, might be at the basis of the unusual reactivity of the HgC bond. As a matter of fact, a natural route toward HgC detoxification involves the action of a bacterial enzyme whose active center is thought to have a tripodlike coordination site schematically represented as4
E:;
BH*
(B=Base)
Several theoretical works dealing with Hg compounds have been published during the past 10 years. The use of semiempirical methods, in particular MND05 and Ah41,6 is wellestablished, and good parameter sets are now a~ailable.~ They are currently being used in a theoretical study of the kinetics of HgC bond cleavage in methylmercury compounds.* Several Universith Frederico 11. f Universitk Degli Studi di Firenze. Abstract published in Advance ACS Abstracts, July 15, 1995. +
@
other works are based either on density functional theory (DFT)9-'' or on Hartree-Fock (HF) and post-HF models."-I8 The importance of relativistic effects has been stressed repeatedly, for both approaches. It appears now well established that these effects can modify bond lengths by about 0.2 A and significantly affect also energetic q~antities.~,' Mercury has a total of 80 electrons, and, as a consequence, it has been generally treated with methods which avoid explicit consideration of inner electrons. This is possible using either frozencore basis sets9 or pseudo potential^.'^^'^ As a matter of fact, all electron calculations have only been performed in order to proof the accuracy of pseudopotentials and to estimate the error introduced by their use.I3 Most theoretical studies of mercury have been devoted to hydrides'2b~12c*'3b,13d of mercury and to linear HgX2 comp o u n d ~ ~ and, ~ ~in~particular, ~ ~ ~ to, the ~ type ~ , of~ Hg ~ orbitals ~ ' ~ involved in the HgX bonding. The prominent role of s-dZz hybridization has been noticed already in the work of Orgel in 1958,20whereas the involvement of pz orbitals in the bonding has never been fully understood. The study of HgzX2"~'~ compounds and of the direct interaction between mercury atoms has also received attention. On the other hand, only a few theoretical studies have been devoted to compounds containing the HgC bond?e,10,17This is in contrast with the role, noxious as well as useful, of the MeHg+ moiety occumng in nature as well as resulting from human activity.' Motivated by the importance of the HgC bond, we have initiated a series of theoretical studies,8 of which the present publication provides the basis. We present here an ab-initio study on a series of HgMe+ containing molecules. In a first stage, we have selected a suitable basis set through calculation of geometries, vibrational frequencies, and thermochemical parameters of a series of simple compounds for which good experimental data are available. Thereafter, special attention has been focused on the relative stability of the HgC bonds of the molecules in terms of electronic structure and of general trends derived from energetic considerations. 1,13317*19
11. Computational Details Ab initio calculations have been performed with the Gamess program obtained from Schmidt et al.21 Except when explicitly
0022-3654/95/2099-12743$09.00/0 0 1995 American Chemical Society
12744 J. Phys. Chem., Vol. 99, No. 34, 1995
Barone et al.
indicated, the core electrons were replaced by the relativistic pseudopotentials known as SBK.'2dq22 The valence basis sets were taken from the same references. In order to have the same number of primitive Gaussian functions in the basis set for all the halogen atoms, a fifth Gaussian function with exponent 0.037 54 was added to the (3,l) valence set of C1. The valence basis sets were contracted with the (3,1,1) pattern. The same exponents were used for s and p functions (sp basis set). The basis set of C, P, C1, Br, and I was completed by a single shell of polarization d functions with the standard exponents provided by Gamess: 0.800,0.550,0.750,0.389, and 0.266, respectively. Following the results of ref 23, a further shell o f f polarization functions with exponent 0.305 was added to the valence set of Hg. The complete basis set used is atom
pseudopotential
contraction
[-I
H C
(31) (3111) (31/1) (31 1/1) (311/1) (31 1/1) (51 11/31 1/1)
[He1 [Ne1 [Ne] [Ar 3dI0] [Kr 4dI0] [Kr 4dI04fl4]
P C1 Br
I Hg
TABLE 1: Experimental Distances of the HgX2, MeHgX, and Me& Compounds ~~
comuound
point group
+
-.
AHr = H(A)
+ H ( B ) - H(AB)
In (1) Eel is the electronic energy of the isolated molecule computed after geometrical optimization, Ezp is the zero point energy, and the other terms take into account the energy of vibrations, rotations, and translations at finite temperature (298 K throughout this work).26 The same principles have been applied for calculating bond dissociation energies and standard
~
_
_
HgCl = 2.27,29 2.25230 HgBr = 2.M29 HgI = 2.61,29 2.55431 HgCl = 2.28232 HgC = 2.06132 HgBr = 2.40632 HgC = 2.07432 HgI = 2.52833 HgC = 2.08733 HgC = 2.08334
D-h D-h D-h c 3Y
c3" c 3"
D3h, D3d
heats of formation. Experimental corrections taken from ref 28 have been applied, when needed, in order to compare the heat of formation computed in the gas phase with the available experimental data obtained in condensed phases. Note that the above procedure is not strictly applicable for MeHg(PH3)3 since the Hg(PH3)3 fragment only represents a convenient model for the actual complex. The energy of interaction between the Me group and the LHg' moiety has been computed from the following reaction: LHgCH,
The influence of the size of the basis set for C on the geometry optimizations was checked performing calculations with the allelectron 6-31G basis set?3 in some cases augmented with one d polarization function (6-31G* basis set).24 In order to evaluate the quality of the chosen basis sets, we have computed observables for a series of simple compounds of formula HgX2, MezHg, and MeHgX (X = C1, Br, I) for which experimental data are available in the literature. The following properties have been calculated: geometry, vibrational frequencies, standard heat of formation, bond dissociation energies, and heat of reaction. The geometry of MeHg(PH3) was also optimized in order to check the quality of the basis set of P. The calculations for [RHg(np3)]+ (R = Me, X) have been performed on model complexes of formula RHg(PH3)3, in which three phosphines replace the bulky np3 ligand. This approximation has already been used with success to model the [(cp3)Cop31 system (cp3 = l,l,l-tris(diphenylphosphinomethy1)ethane),25 and its validity is further discussed in some detail later. Calculations have been performed at the HF level and also including correlation at the second order in perturbation theory according to the Mraller and Plesset formalism (MP2).26927They will be simply indicated as HF and MP2 or HF+f and MP2+f when f functions on Hg have been also included. A. Geometry Optimizations. Geometry optimizations have been performed using energy-only optimization techniques, since the current version of Gamess2' does not compute energy gradients in the presence of f functions. B. Energetic Calculations. The enthalpy of a reaction A B AB (Mr)is calculated assuming an ideal gas behavior according to
~
distances (A)
EcHg= E(LHg)
+
LHg'
+ CH,'
+ E(CH3) - E(LHgCH3)
(2)
The geometry of the LHgCH3 molecule has been optimized and the atomic coordinates retained in the calculation of the separated fragments. The energies computed in this way can be regarded as fragment interaction energies. The comparison with dissociation energies (when available) is discussed in the text. C. Vibrational Frequencies. The calculation of harmonic vibrational frequencies requires the second derivatives of the energy with respect to the atomic positions.26 Since these derivatives are not available in the current version of Gamess in the presence o f f functions in the valence basis or at the MP2 level, vibration frequencies have been calculated at the HF level, without f polarization functions on Hg. A similar procedure has been applied with success in a recent study of HgX2 and other
III. Results and Discussion A. HgXz, MeHgX, and MezHg Compounds. In order to choose a basis set for Hg, which allows one to handle large molecular systems with reasonable accuracy, we have performed calculations on simple molecular systems of formulas HgX2, MeHgX, and Me2Hg. On these systems, calculations can be done at almost any level of approximation and the accuracy of the model in reproducing experimental observables can thus be checked. In particular, geometry optimizations have been performed and vibrational frequencies and thermodynamic properties calculated. In the remaining of the paragraph, we will compare results obtained with different basis sets and methods of calculation. Of course, since our future aim will be that of handling large molecules which can mimic naturally occurring organomercury compounds, more rigorous methods such as extended configuration interaction (CI) or inclusion of all electrons have not been pursued for. The available experimental geometries of our test compounds are collected in Table 1. Geometry optimizations have been performed within the point groups indicated in the table. Two geometrical arrangements have been considered for Me2Hg: the staggered conformation (& symmetry) and the eclipsed one (D3h ~ y m m e t r y ) . ~In~the . ~ ~calculations, the CH distances and
_
_
J. Phys. Chem., Vol. 99, No. 34, 1995 12745
Hg-C Bonding of Methylmercury(I1) Compounds
TABLE 2: Distances (A) Obtained from Geometry Optimization@ (a) At Various Levels of Approximation
compound
distance
HF
HFff
MP2
MP2+f
2.327 (2.5; 3.3) 2.453 (0.53) 2.641 (1.2; 3.4) 2.375 (4.0) 2.129 (3.3) 2.490 (3.5) 2.139 (3.1) 2.675 (5.8) 2.150 (3.0) 2.161 (3.7)
2.310 (1.8; 2.6) 2.441 (0.04) 2.627 (0.65; 2.8) 2.354 (3.2) 2.128 (3.3) 2.478 (3.0) 2.132 (2.8) 2.656 (5.1) 2.140 (2.5) 2.155 (3.5)
2.305 (1.5; 2.4) 2.435 (-0.2) 2.624 (0.54; 2.7) 2.344 (2.6) 2.101 (1.9) 2.468 (2.6) 2.110(1.7) 2.656 (5.1) 2.121 (1.6) 2.136 (2.5)
2.263 (-0.31; -0.49) 2.400(-1.6) 2.578 (1.2; 0.94) 2.302 (0.88) 2.084 (1.1) 2.436 (1.2) 2.091 (0.82) 2.605 (3.0) 2.099 (0.57) 2.116 (1.6)
(b) Using Various Basis Sets for Carbon
compound MeHgBr Me2Hg
distance HgBr HgC HgC
SBKb
6-3 1Gb
SBK*‘
6-3 1G*‘
2.436 (1.2) 2.098 (1.2) 2.125 (2.0)
2.436 (1.2) 2.088 (0.68) 2.114(1.5)
2.436 (1.2) 2.090 (0.77) 2.116 (1.6)
2.436 (1.2) 2.082 (0.39) 2.107 (1.2)
Values in parentheses refer to percentual error with repect to the experimental data reported in Table 1 (see text: eq 3). No d polarization function on carbon. One d polarization function on carbon. (I
the HCH angles of the methyl groups have been fixed at 1.099 A and 109.lo, respectively. Complete geometrical optimizations, including the optimization of the methyl groups, have been performed on Me2Hg and MeHgBr. Fully relaxed and partially frozen computed geometries are identical within experimental error, supporting our simplified approach. The results of the geometry optimizations are collected in Table 2. In Table 2a the results of HF and MP2 methods are compared. The effects of f polarization functions on Hg are also shown. The numbers in brackets are computed as
TABLE 3: Vibrational Frequenciet? Computed at the HF Level for HgX2, MeHgX, and MeZHgb HgCh HaBrz Hd2 48 91 61 6(XHgX) 226 (238)’ 392 (408)’ 280 (293)’ v, (HgX) nu 341 (353)’ 212 (224)’ 153 (164)’ vs (HgX)
xu+ c.+
vi (HgX) V, (HgC) 6.js(CH3) V, (CH3)
6(C-Hg-X)
(3)
e (CH3)
6da (CH3) V, (CH3)
u corresponds to a percentual error on the calculated values. When two values are given, they refer to the different experimental data available in the literature. The calculated bond distances are generally overestimated, the presence of the f function as well as the use of MP2 improving the agreement with the experiment. In general, the MP2+f level of calculation gives the best results, with errors not larger than 1.6%. The only exception is MeHgI in which the error for the HgI distance rises to 3%, at best. The simplest level of computation (HF) gives errors on the HgC and HgX bond distances lower than 4% (with the exception of HgI), also in reasonable agreement with experiment. Optimizations of the geometry of HgMe2 have been performed in D3h (shown in Table 2a) as well as D3d Symmetry. The former calculation reproduces the experimental geometries more accurately and provides the lowest total energy. This result agrees with previous studies in which IR and Raman spectra have been assigned within this ~ y m m e t r y . ~Only ~ . ~D3h ~ symmetry has been used in the other computations. Further calculations have been performed on MeHgBr and HgMez, using SBK and 6-31G or 6-31G*basis sets for C. The results of these calculations are shown in Table 2b. The best agreement is obtained with the 6-31G basis set, but the improvement with respect to SBK is only 0.006-0.008 A. Adding the d polarization functions to both basis sets also improves the result only by 0.008-0.01 1 A. We conclude that the use of SBK with one polarization function provides the best compromise between computational time and accuracy. The computed vibrational frequencies are collected in Table 3. The CH3 vibrations have been scaled by 0.9, as generally
AI AI AI AI E
E E E
MeHgCl
MeHgBr
MeHgI
333 (336)d 540 (554)d 1191 (1196)d 2859 (2925)d 115(135)d 761 (805)d 1414 (1415)d 2965 (301 l)d
230 (228)d 531 (545)d 1185 (1190)d 2859 (2925)d 105(121)d 753 (798)d 1413 (1415)d 2965 (3009)d
181 (181)d 522 (533)d 1179(1177)d 2859 (2921)d 99(112)d 744 (786)d 1412 (1410)d 2965 (3008)d
Me2Hg v,(CHg) 6, (CH3) v S (CH)
AI AI A] AI”
MezHg d,(CHgC)
498(515)’ 1156 (1 182)’ 2950 (2910)’ 105 (-) 532 (550)‘ 2938 (2880)’ 2938(2880)’
er(CH3)
6 d (CH3)
E’ E’ E’ E‘ E” E E”
140(156)’ 740 (787)’ 1410 (1475)’ 2955 (2966)’ 659 (700)’ 1408 (1443)’ 2847(2869)’
v(CH) er(CH3) (CHg) AT 6 d (CH3) AT 6 d (CH3) v(CH) AT v(CH) a Notations: v = stretching; 6 = bending; t = torsion; e = rocking; a = antisymmetric; s = symmetric; d = deformation. Experimental values in parentheses. From ref 37. From ref 38. e From refs 35 and t V,
36.
performed in literature,26 while the vibrational frequencies involving Hg have not been corrected. The agreement with the experimental data is satisfactory. The Hg vibrational frequencies are generally slightly underestimated, but never by more than 20 cm-I. This is, naturally, somewhat less accurate than results from empirical fittings based on the GF matrix method35,36,38 but seems largely sufficient for structural identification. Computed thermodymamic properties are compared to experimental data in Tables 4a,b and 5. Bond dissociation energies computed at the MP2+f level are shown in Table 4b. D I and D2 correspond to the first and second dissociation energies respectively. For compounds of the type MeHgX
-
+ CH,’ HgX -Hg + X
CH,HgX
Dl
D2
‘HgX
(4)
12746 J. Phys. Chem., Vol. 99, No. 34, 1995
Barone et al. of formation computed in this way and corrected to condensed phase (see section IIb) are shown in Table Sa. The agreement with the experimental values (in parentheses) is reasonable for HgX2 but really bad for MeHgX. The largest error, in the last cases, comes probably from the heat of formation computed for CH3. This is confirmed by calculating the heat of formation from the reaction
TABLE 4: Dissociation Energies (kcaVmolp of HgX2, MeHgX, and MetHg comoound
Di (a) Calculatedat 52.8 (81) 48.3 (72) 43.1 (61) 34.3 (64.3) 32.6 (61.8) 29.4 (59.0) d(51.5)
HgCh HgBn HgI2 MeHgCl MeHgBr MeHgI MezHg
D
0 2
the HFb Level 16.0 (24) 12.5 (17) 8.8 (8) 16.2 (24) 12.5 (17) 8.8 (8) d (6.9)
68.8 (106) 60.7 (89) 5 1.9 (69) 40.5 (88.3) 45.1 (78.8) 38.2 (67) d (58.4)
(b) Calculated at the MP+fc Level 83.1 [82.3] 21.4 [20.6] 74.7 [71.2] 17.4 [13.9] 67.1 [59.8] 12.6 [5.40] 68.9 21.4 [20.6] 65.5 17.4 [13.9] 63.0 12.6 [5.40] 63.0' -0.875'
HgCh HgBr2 HgI2 MeHgCl MeHgBr MeHgI Medg
CH3
92.1 [85.1] 79.7 [65.2] 90.3 [89.5] 82.9 [79.4] 75.6 [68.4] 62.1
(CH,),Hg
+
Dl, D2: first, second dissociation energy (see text). D = DI D2. Experimental data, taken from ref 39, are given in parentheses. Figures in square brackets have been corrected for atomic spin-orbit coupling contributions of the halogen^.'^' Optimization of the fragment MeHg in HF failed. The corrective terms to E,( in (1) for the MeHg' fragment have been calculated on the MP2+f geometry.
+ HgX, - 2CH3HgX
(6)
The heat of formation of MeHgX, A@, can be computed using (7), once the energy of the reaction, AHr, has been computed and if the experimental values for the formation enthalpies of Me2Hg and HgX2 are available. Equation 7 can, of course, be Mr
= 2@(CH3HgX)
- @((cH3)2Hg)
- @(HgX2) (7)
TABLE 5 (a) Heats of Formation (kcal/mol) of HgX2, MeHgX, and MezHg, Computed at the MP2Sf Level" comoound HgCh HgBn HgI2 MeHgCl MeHgBr MeHgI MezHg
(5)
The results of these calculations are collected in the second column of Table Sa. Inclusion of the spin-orbit coupling corrections does not ameliorate the agreement with the experimental data. The third reaction used is the isodesmic reacti~n:~~
104.5 [102.9]
a
+ X + Hg - CH3HgX
AHpo(elem)b
AHq'(CH2Y
-31.8 [-30.2Id (-35.O)e -23.8 [-16.8Id (-20.4)e -14.1 [O.418ld(-4.1)' 34.9 [35.7Id(-12.5)e 39.8 [43.3Id(-4.4)' 46.1 [53.3Id(5.2)' 114.3 (22.5)'
-9.24 [-8.36Id (-12.5)' -4.4 [-0.89Id (-4.4)' 3.4 [1O.6ld(5.2)' 26.0 (22.5)'
(b) Heats of the Isodesmic Reaction 6, (AHr), and Related Heats of Formation (kcaVmo1) of HgXz, MeHgY, and MezHg, Computed at the MP2+f Level" C1 -13.7 (-14)' Br -11.4 (-l0)f I -8.5 (-8)'
-33.8 (-35)g -19.9 (-20.4)g -3.6 (-4.lY
23.7 (22.5)g 23.0 (22.5)g 23.0 (22.5)g
-13.1 (-12.5)s -4.7 (-4.4)g 5.0 (5.2)g
a Experimental values given in parentheses. From single atoms. From eq 5. Spin-orbit corrections on halogen atoms included. e Reference 7b. f Reference 39.
Calculations have also been performed at the HF level with and without f polarization functions and at the MP2 level without f functions on Hg. The results of the HF calculations are shown in Table 4a. The agreement with the experimental data is definitely worse. For HgBrz, for example, the following results are obtained: (1) HF, 60.8 kcdmol; (2) HF+f, 64.8 kcaymol; (3) MP2,92.9 kcaYmo1, to be compared with experimental and MP2+f values of 89 and 92.1 kcdmol, respectively. On these grounds we have performed all the other energy computations at the MP2+f level. Corrections for the spin-orbit coupling energies of the halogen atoms, following the procedure outlined in ref 15c, have been also applied. Significant improvement of the computed values was observed in the case of iodine containing compounds. According to the procedure outlined in ref 15c, the molecular fragments and the Hg atom have not been corrected for spin-orbit coupling. For the calculation of the standard heat of formation we have applied (1) to several possible reactions which lead to the same molecule. The first reaction is the formation of the molecule from the individual atoms in the gas phase. The standard heats
used to compute any of the enthalpies appearing on the right hand side, when the enthalpies of the other two species are known. Applying this procedure in turn, we have computed the standard enthalpies of formation, reported in Table 5b. The agreement with the experimental data is very good, thus giving further support to the practice of using isodtsmic reactions for the computation of heats of formation. The above results show that the MP2 calculations are, in general, in better agreement with the experimental data than HF ones. This is due to the inclusion of correlation energy in MP2, even if as a perturbation. On the other hand, HF is still useful, if one is only interested in relative variations or specific trends of the experimental observables. For example, the bond dissociation energies, D,calculated at the HF level (Table 4a) correctly decrease in the order C1 > Br > I. B. [MeHg(PH&]+. The number of mercury-phosphine complexes bound to carbon reported so far is limited, and a complete structural determination has only been done for [PhHg(PPh3)]+ (Ph = CsH5).40 The HgP and HgC distances were found to be 2.431 and 2.09 A. In other monophosphine complexes4' of the type [X2Hg(PPh3)]n (X = C1, Br, I; n = 1, 2) the HgP distances are in the range 2.36-2.46 A, whereas a wider range of HgP distances (between 2.45 and 2.91 A) has been reported42for diphosphine complexes of the type [HgX2(PPh3)2] (X = CN, NO3, C1, Br, I). The geometry of [MeHgPH# has been optimized in Cjv symmetry. The computed HgC distance (2.082 A) is very close to that of [PhHg(PPh3)]+,and the computed HgP distance (2.519 A) falls in the range of the observed distances, thus supporting the choice of the basis set for P. Complexes of the general formula [RHg(np3)]+ (R = Me, ..., C1, Br, I) have been recently prepared and the crystal structure of the Me3 and 143derivatives determined. The most significant geometrical parameters of the Me derivative are shown in Figure 1. Although the whole molecule does not possess any crystallographic symmetry, the local environment of Hg is very near a C, symmetry. A more regular structure has been observed in the I derivative, shown in Figure 2, whose symmetry approaches the CJ" limit of a trigonally distorted pseudotetrahedron. In order to perform calculations on RHg(np3) systems with the
J. Phys. Chem., Vol. 99, No. 34, 1995 12747
Hg-C Bonding of Methylmercury(I1) Compounds
TABLE 6: Distances and Angles of the Model Compounds [XH~(PHJ)~]+, Optimized at the MP2+f Level X Hg-P (A) Hg-X 6) P-Hg-X (deg)
c1 Br I
Figure 1. Experimental geometry of the [CH3Hg(np3)]+compound. Distances are in angstroms.
11
Figure 2. Experimental geometry of the [IHg(np3)]+ compound. Distances are in angstroms.
Figure 3. Optimized geometry of the model compound [CHlHg(PH3)3]+. Distances are in angstroms.
minimum requirement of computer resources, we have replaced the tripod np3 ligand by three phosphine groups. This model allowed a successful rationalization of the electronic properties and reactivity of the (cps)Co(P3) complex (cp3 = l,l,l-tris(diphenylphosphinomethy1)ethane) moiety using a density functional approach.25 The geometry of the model complex [MeHg(PH3)3]+ has been optimized in C, symmetry, and the results of the calculations are graphically shown in Figure 3. During the optimization procedure the eometry of CH3 and PH3 groups were fixed [C-H = 1.099 , C-H-C = 109.1", P-H = 1.440 A, H-P-H = 93.4'1. In the optimized geometry the three HgP distances are equal and the final symmetry is very near C3". The agreement between the computed geometrical parameters and those observed in the [MeHg(np3)]+ complex is good. The largest deviation from experimental data concerns the CHgP angles which are identical (123') in the
x
2.51 2.58 2.60
2.31 2.50 2.65
109.6 110.6 113.6
model compound, whereas in the real molecule they are smaller and significantly different from each other (121.6(9)", 121.5(9)', and 111.0(9)'). In order to investigate the effect of structural deformations on the computed geometries and to the effect of constraints on the optimization procedure, we have also performed two constrained geometry optimizations. In the first one (case 1) the H P bond distances have been fixed at 2.808,2.605, and 2.605 and the CHgP angles at 11lo, 121.6', and 121.6", according to the experimental results (average values have been used for two nearly identical groups). In the second optimization (case 2) only the CHgP angles were fixed. The results are Hg-C = 2.153 8, for cases 1 and 2, respectively, ahd Hg-P1 = 2.696 A, Hg-P2 = 2.656 A for case 2. The agreement with the experimental data is apparent in both cases. It is interesting to note that the HgC distance is always well reproduced and that the total energies of the molecule in the two cases differ by only 1.88 kcavmol. This suggests that the calculation of relative thermodynamic properties and reaction energies remain meaningful, even using this seemingly drastic model of the true complex. Finally, geometry optimizations have also been performed on the complexes [XHg(PH3)3]+ (X = C1, Br, I). Crystallographic data are available only for the [IHg(np3)]+complex43 and have already been shown in Figure 2. [EIg(PH3)3]+is more symmetric than [MeHg(PH3)3]+,with PHgI angles varying only from 107" to 114' (versus 111' to 122" for PHgC) and with PHg distances ranging between 2.52 and 2.56 8, (versus 2.60 and 2.81 A). Taking also into account that a structure of C3" symmetry was obtained from the complete optimization of [MeHg(PH3)3]+, we have imposed this symmetry in all the other geometry optimizations. The results are collected in Table 6. The computed HgI bond distance is shorter than the experimental value, while all the other geometrical parameters are satisfactory. In particular our computations are able to reproduce the significant decrease of the HgP bond lengths with respect to those observed in [MeHg(PH3)3]+ and the decrease in HgP distance resulting from substitution of C1 by Br or I. This latter result is experimentally well evidenced in compounds of the type X2HgPR34'a and X ~ H ~ ( P R ~ ) Z . ~ ~ ~ C. Electronic Structure. The most common coordination number for Hg is surely 2. However, the ability of Hg to complete its coordination sphere is evidenced, for instance, by its 6-fold coordination in the simple HgX2 compounds in the solid state. In the linear D,h molecules Z-type interactions involve the s, pz, and dZ2orbitals on Hg and s and pz orbitals on X. II-interactions occur between the px, py, dyz,d,, orbitals on Hg and the px, py orbitals of X. The dx2-y2, d, orbitals on Hg remain nonbonding unless A-type orbitals are introduced in the valence shell of X through polarization functions. These symmetry-allowed orbital interactions are summarized in Table 7a. A similar rationalization is easily established for MeHgX and [MeHgPH# molecules (Table 7b), belonging to the Cj" point group. In particular, the orbitals of al symmetry accomplish o-interactions, whereas the orbitals of e symmetry participate to n-and &interactions. The lone-pair orbitals of P atoms in the three phosphine moieties of [MeHg(PH3)# have the same symmetry behavior as the hydrogen s orbitals of PH3. In particular the HOMO (which dominates the interaction with CH3) corresponds to the a1 symmetry-adapted combination.
x
12748 J. Phys. Chem., Vol. 99, No. 34, 1995
Barone et al.
TABLE 7: Schematic Presentation of Symmetry Allowed Orbital Interactions of HgX2 and MeHgX
HgX2, D-h Tir
X2"*b
Hg
&+
s, 22
U
SI
+ s2
P:I - Pz2
C"'
P:
U
n, n,
SI
xz
n
Px,
4
+ P:2
pxl
- p12
px1
-+ px2
p\1 - p\2
YZ
n
- s2
Prl
PI
x2 - Y2, Xy
P,1
+ PI2
MeHgX, C3" a]
u s, p:,
e
n pr. p,, YZ, xz,
z2
p: px, p, S,
~ ( PAC), 0 , si(H) + s2(H) + s 3 W
px(C), p,(C), s2(H) - sdH), 2sdH) -
s2W) - s3(H) The orbitals of the two X atoms are distinguished by subscripts 1 and 2. The functions are not normalized. The orbitals of the three H atoms are distinguished by subscripts 1-3. x2
- y2, xy
a
Our calculations show that, for all the compounds, the HgC bond is primarily stabilized by a-interactions, n-contributions being rather small. In particular, the strongest a-interaction occurs between the SOMO of the methyl fragment and the SOMO of the remaining HgL fragment. It is noteworthy that this interaction and the shape of the resulting MO are very similar in all the compounds, including the pseudotetrahedral [MeHg(PH3)3]+ system. Hg binds to C via its 6s, 6p, as well as 5d,2 orbitals. The resulting MO's form the HOMO of MeHgCl and [MeHg(PHs),,]+and the HOMO-1 in MeHgBr and MeHgI. The HOMO of these last two compounds is essentially a X lone-pair orbital. The hybridization of Hg in linear complexes has been a subject of interest in l i t e r a t ~ r e . ' ~ . 'The ~ . ' ~Mulliken gross orbital populations, presented in Table 8, provide one criterion for deciding on the type of orbtials involved in the Hg-ligand bond. It is apparent that the 6s orbital of Hg provides the most important contribution, whereas the d orbitals contribute only little, or not at all in the case of [MeHg(PH&]+. The Mulliken population analysis indicates a definitely large contribution of Hg(6p) orbitals. For complexes where Hg is in a 2-fold coordination, the 6p, orbital accounts almost completely for the 6p population and points to a sp-like hybridization. In the pseudotetrahedral triphosphine complex, significant contributions also arise from px and py orbitals, thus establishing some interaction with the three phosphorous atoms. Recently, the NBO methodMwas used to describe the bonding in HgX2 showing that the coordination of Hg is accounted for by the 6s orbitals and the 6p orbitals do not appreciably participate to the bond. We have performed a similar analysis on the methylmercury compounds. The results are shown in Table 8. They show that mercury essentially uses the 6s orbital to form the bonds, the contribution of the 5d and 6p orbitals being negligible. Note that s orbitals alone could support a pseudotetrahedral coordination of Hg with the tripod np3 ligand in view of repulsions between different electron pairs and the rigid structure of the ligand. It is apparent that a definite answer on the nature of the Hg orbitals involved in the formation of complexes cannot be given, since different decomposition procedures of the molecular electronic density give contradictory results. Both Mulliken and NBO analysis indicate, in any case, that the 6s orbitals provide the strongest contribution. The negligible role of p orbitals suggested by the NBO method is, in our opinion, questionable
and could derive from the limits of a localized orbital picture to describe the bonding of mercury. D. Relative Stability of the HgC Bond. We are especially interested in the stability and reactivity of the methylmercury compounds, which are at the basis of environmental problems caused by mercury. Of course, definite conclusions about reactivity can only be drawn from a knowledge of the different transition states. Already in the present stage we have, however, several criteria which provide a hint to the reactivity and stability of the MeHg entity, based on thermodynamics and electronic properties. The principles following from the frontier orbital theory, can be used as a fist indication of the relative reactivities of the compounds, provided that the virtual orbitals are well isolated from the HOMO. This is true for the present molecules. The decrease of the HOMO-LUMO gap suggests that reactivity increases in the following order: C1 < Br < I; n = 1 < n = 3. Moreover the HgC overlap population in the HOMO (HOMO-1 in the case of MeHgBr and MeHgI) decreases in the order C1 > Br > I for MeHgX and from [MeHg(PH3)]+ to [MeHg(pH&]+. This might point to a decrease of the HgC bond strength in the same order. The NBO analysis (Table 8, last column) provides a strong donor-acceptor interaction, which actually accounts for the HgX or HgP bond stabilization and a HgC bond destabilization, in the order C1, Br, I and n = 1, n = 3. The interaction energies between the fragments CH3* and 'HgL, defined in eq 2, are as follows (kcavmol): MeHgX: 78.8 for X = C1 > 75.9 for X = Br > 73.2 for X = I
80.6 for n = 1 > 70.6 for n = 3
[MeHg(PH3),]:
The interaction energies of the MeHgX series are lower than the dissociation energies shown in Table 4b. This is because the latter are derived from fragments whose geometries have been optimized. However, differences are small and, what is more important, both types of energies correctly decrease within the MeHgX series according to the sequence C1 > Br > I. This suggests that interaction energy can be taken as good estimates of the relative HgC bond strengths within a series of related molecules. Consequently, the pronounced decrease in interaction energy of the [MeHg(PH3),]+ (n = 1, 3) complexes from the monophosphine to the triphosphine compound can be explained as due to the weaker HgMe bond of the latter. Recently, the reactivity of [(np3)HgR]+ (R = methyl, phenyl) with CH3COOH acid has been studied e~perimentally.~~ The phenyl derivative reacts almost quantitatively at room temperature, while the reaction of [(nps)HgMe]+ proceeds with glacial acetic acid at 40 "C. In the same conditions, RHgX compounds do not react but rather require high temperatures and strong mineralic acids like HCl, HC104, and H Z S O ~ .In~ order to rationalize these observations from a thermodynamical point of view, we have calculated the heat of the following reactions:
+ HX -HgX, + CH4
(8)
+ HX
(9)
CH3HgX CH-,Hg(PH,),
-
XHg(PH3)3
+ CH4
The geometries of the various molecules have been optimized, but no corrections are applied for vibrations, translations, and rotations. The heats of reaction vary as follows (kcavmol): reaction(8): -23.3 for X = C1 < -29.3 for X = Br
4
-36.7 for X = I
reaction(9): -26.7 for X = C1 < -33.0 for X = Br < -39.6 for X = I
J. Phys. Chem., Vol. 99, No. 34, 1995 12749
Hg-C Bonding of Methylmercury(I1) Compounds
TABLE 8: Mulliken and NBO Gross Populations for Hg, NBO Charges, and Donor-Acceptor Second Order NBO Energies (kcallmol) Hg Mulliken population Hg NBO population NBO charges donor-acceptor energy compound 6s 6p 5d 6s 6p 5d C CH3 Hg X" X-HgC or P-HeC MeHgCl MeHgBr MeHgI [MeHgPH3+]+ [MeHg(PH3)3]+
1.16 1.03 1.04 1.13 0.95
0.59 0.59 0.62 0.52 0.82
9.93 9.91
9.93 9.91 9.99
0.86 0.87 0.89 0.93 0.72
0.04 0.04 0.04 0.03 0.07
9.88 9.89 9.90 9.90 9.94
-1.06 -1.07 -1.08 -1.00 -1.10
-0.49 -0.50 -0.51 -0.37 -0.52
1.23 1.19 1.15 1.14 1.26
-0.73 -0.70
-0.64
114 119 130
0.23 0.26
95
88
X for MeHgX, P for [MeHg(PH3),Ii
These results are in agreement with experimental observations. Firstly, both reactions are thermodynamically allowed. Secondly, reaction 9 is easier than reaction 8: this is in agreement with the fact that the [(np3)HgMe]+ complex can react also in weak acidic conditions, while MeHgX requires strong acids. The trend observed for reaction 8 on changing X is also in agreement with the HgC interaction energies calculted previously and shows that, thermodynamically,reactions proceed with increasing ease in the order C1 < Br < I. The charges of the CH3 group and of Hg in the reagent could be further indicators of the reactivity order toward HX. Attack, of the electrophile H+ will be more favorable if C (CH3) is more negatively charged. This occurs in the order PH3, C1, Br, I, (PH3)3. On the basis of the charge of Hg, one might also conclude that the affinity of the positively charged Hg for Xis largest for the triphosphine complex. All the arguments point toward the same conclusions: the HgC bond cleavage occurs with increasing ease in the order [MeHg(PH3)]+, [MeHg(PH3)3]+on the one hand and MeHgCl, MeHgBr, MeHgI on the other hand. On the basis of the lowest HgC bond interactions, the more negative energy of reaction 9 with respect to reaction 8 and the most favorable charge of C and Hg, we also conclude that the cleavage of the MeHg bond by HX is easiest in [MeHg(PH&]+.
Conclusions In this study we have investigated the role of substituent effects in determining the nature and the strength of the HgC bond. From a methodological point of view medium size basis sets provide reliable general trends for a number of properties already at the HF level. However, quantitative results can only be obtained including correlation energy (even at the MP2 level) and adding f polarization functions on Hg. Inclusion of spinorbit corrections on halogen atoms has resulted in a fundamentally better agreement in the case of total dissociation energies. At this level geometrical parameters are sufficiently accurate for most purposes, and reliable thermochemical data can be obtained by using isodesmic reactions. From a more general point of view, all the computations indicate that HgC bond cleavage is favored decreasing the electronegativity of further substituents or going from linear to pseudotetrahedral coordinations of Hg. In this connection it is particularly significant that all the molecular indicators (HOMO-LUMO gap, HgC overlap populations, atomic charges on the Hg and CH3 fragments) point in the same direction. In summary it seems quite safe to say that the route is paved for reliable theoretical studies of larger biologically significant methylmercury compounds and of the kinetics of acid cleavage reactions.
Acknowledgment. Thanks are expressed to Prof. P. Fantucci and Dr. S. Midollini for many discussions and helpful suggestions. M.G.U. acknowledges the European Community for a research grant. Thanks are expressed to one of the reviewers
for his comments on the inclusion of spin-orbit coupling corrections to the dissociation energies.
References and Notes (1) (a) Wood, J. M.; Kennedy, F. S.; Rosen, C. G. Nature 1968, 220, 173. (b) Nelson, N. Environ. Res. 1971,4, 1. (c) Bache, C. A,; Gutenmann, W. H.; Lisk, D. J. Science 1971, 172, 951. (d) Summers, A. 0.;Silver, S. Annu. Rev. Microbiol. 1978, 32, 637. (e) Jensen, S.; Jemelov, A. Nature 1969,223,753. (f) Friberg, L. T.; Vostal, J. J.; Mercury in the Environment; CRC Press: Cleveland, 1972; p 232. (g) Sheldrick, W. S.; Heeb, G. lnorg. Chim. Acta 1992, 194, 67. (h) Camargo, J. A. Nature 1993, 365, 302. (i) Gardner, M. J.; Gunn, A. M. Nature 1993, 366, 118. (j)Darbieu, M. H. Rev. Roum. Chim. 1993, 38, 219. (k) Vimy, M. J. Chem. lnd. 1995, 14. (2) (a) Kreevoy, M. M. J . Am. Chem. Soc. 1957,20,5927. (b) Keevoy, M. M.; Hansen, R. L. J . Am. Chem. SOC.1961, 83, 626. (3) (a) Ghilardi, C. A.; Innocenti, P.; Midollini, S.; Orlandi, A.; Vacca, A. J . Chem. SOC.,Chem. Commun. 1992, 1691. (b) Barbaro, P.; Cecconi, F.; Ghilardi, C. A.; Midollini, S.; Orlandi, A,; Vacca, A. lnorg. Chem. 1994, 33, 6163. (4) Moore, M. J.; Distefano, M. D.; Zydowsky, L. D.; Cummings, R. T.; Walsh, C. T. Acc. Chem. Res. 1990, 23, 301. (5) Dewar, M. J. S.; Thiel, W. J . Am. Chem. SOC. 1977, 99, 4899. (6) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J . Am. Chem. SOC. 1985, 107, 3902. (7) (a) Dewar, M. J. S.; Grady, G. L.; Merz, K. M.; Stewart, J. J. P. Organometallics 1985.4, 1965. (b) Dewar, M. J. S.; Jie, C. Organometallics 1989, 8, 1547. (8) Unpublished results, work in progress. (9) (a) Snijders, J. G.; Baerends, E. J. Mol. Phys. 1979, 36, 1789. (b) Snijders, J. G.; Baerends, E. J.; Ros, P. Mol. Phys. 1979, 38, 1909. (c) Ros, P.; Snijders, J. G.; Ziegler, T. Chem. Phys. Lett. 1980, 69, 297. (d) Ziegler, T.; Snijders, J. G.; Baerends, E. J. J . Chem. Phys. 1981, 74, 1271. (e) DeKock, R. L.; Baerends, E. J.; Boerrigter, P. M.; Hengelmolen, R. J . Am. Chem. SOC. 1984, 106, 3387. (10) (a) Tse, J. S.; Bancroft, G. M. Creber, D. K. J . Chem. Phys. 1981, 74, 2097. (b) Sano, M.; Adachi, H.; Yamatera, H. Bull. Chem. SOC.Jpn. 1982, 55, 1022. (c) Bice, J. E.; Tan, K. H.; Bancroft, G. M.; Yates, B. W.; Tse, J. S. J . Chem. Phys. 1987, 87, 821. (1 1) Schwerdtfeger, P.; Boyd, P. D. W.; Brienne, S.; McFeaters, J. S.; Dolg, M.; Liao, M.4.; Schwarz, W. H. E. lnorg. Chim. Acta 1993, 231, 233. (12) (a) Basch, H.; Topiol, S. J . Chem. Phys. 1979, 71,802. (b) Bemier, A.; Pelissier, M. Chem. Phys. 1986, 106, 195. (c) Dolg, M.; Kuchle, W.; Stoll, H.; Preuss, H.; Schwerdtfeger, P. Mol. Phys. 1991, 74, 1265. (d) Stevens, W. J.; Krauss, M.; Basch, H.; Jasien, P. G . Can. J . Chem 1992, 70, 612. (13) (a) Hay, P. J.; Wadt, W. R.; Kahn, L. R.; Bobrowicz, F. W. J . Chem. Phys. 1978.69.984. (b) Stromberg, D.; Gropen, 0.;Wahlgren, U. Chem. Phys. 1989, 133, 207. (c) Andrae, D.; Haussemann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor. Chim. Acta 1990, 77, 123. (d) Kiichle, W.; Dolg, M.; Stoll, H.; Preuss, H. Mol. Phys. 1991, 74, 1245. (e) Haussemann, U.; Dolg, M.; Stoll, H.; Preuss, H.; Schwerdtfeger, P.; Pitzer, R. M. Mol. Phys. 1993, 78, 1211. (14) Kaupp, M.; von Schnering, H. G. Angew. Chem., Int. Ed. Eng. 1993, 32, 861.
(15) Kleier, D. A.; Wadt, W. R. J . Am. Chem. SOC. 1980, 102, 6909. (b) Kaupp, M.; von Schnering, H. G . lnorg. Chem. 1994, 33, 4179. (c) Kaupp, M.; von Schnering, H. G. lnorg. Chem. 1994, 33, 2555. (16) (a) Wadt, W. R. J. Chem. Phys. 1980, 72,2469. (b) Schwerdtfeger, P.; Heath, G. A.; Dolg, M.; Bennett, M. A. J . Am. Chem. Soc, 1992, 114, 7518. (c) Howard, S. T. J . Phys. Chem. 1994, 98, 61 10. (d) Akesson, R.; Persson, I.; Sandstrom, M.; Wahlgren, U. Inorg. Chem. 1994, 33, 3715. (17) Schwerdtfeger. P. J . Am. Chem. SOC.1990, 112, 2818. (18) Norrby, L. J. J . Chem. Educ. 1991, 68, 110. (19) Pyykko, P. Chem. Rev. 1988, 88, 563. (20) Orgel, L. E. J . Chem. SOC. 1958, 4186.
12750 J. Phys. Chem., Vol. 99, No. 34, 1995 (21) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.; Gordon, M. S.;Jensen, J. H.; Koseki, S.;Matsunaga, N.; Nguyen, K. A.; Su,S.J.; Windus, T. L.; Dupuis, M.; Montgomery, J. A. J . Comput. Chem. 1993, 14, 1347. (22) Stevens, W. J.; Basch, H.; Krauss, M. J. Chem. Phys. 1984, 81, 6026. (23) Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972, 56, 2257. (24) Hariharan, P. C.; Pople, J. A. Theor. Chim. Acta 1973, 28, 213. (25) Bencini, A.; DiVaira, M.; Stoppioni, P.; Uytterhoeven, M. G. Theor. Chim. Acta 1995, 91, 187. (26) Hehre, W. J.; Radom, L.; Schleyer, P.v.R.; Pople, J. A. Ab-initio Molecular Orbital Theory; John Wiley & Sons: New York, 1986. (27) Dupuis, M.; Chin, S.; Marquez, A. In Relativistic and Electron Correlation Efiecrs in Molecules; Malli, G.,Ed.; Plenum Press: New York, 1994. (28) Barin, I. Thermochemical Data of Pure Substances; VCH: Cambridge, UK, 1993. (29) Allen, P. W.; Sutton, L. E. Acta Crystallogr. 1950, 3, 46. (30) Kashiwabara, K.; Konaka, S.;Kimura, N. M. Bull. Chem. SOC.Jpn. 1973, 46, 410. (31) Spiridonov, V. P.; Gershikov, A. G.; Butayev, B. S.J . Mol. Struct. 1979, 52, 53. (32) Gordy, W.; Sheridan, J. J. Chem. Phys. 1954, 22, 92. (33) Iwasaki, N. Bull. Chem. SOC.Jpn. 1976, 49, 2735.
Barone et ai. (34) Kashiwabara, K.; Konaka, S.;Iijima, T.; Kimura, N. M. Bull. Chem. SOC. Jpn. 1973, 46, 407. (35) Bakke, A. M. W. J . Mol. Spectrosc. 1971, 38, 214. (36) Gutowsky, H. S. J . Chem. Phys. 1947, 17, 128. (37) Givan, A.; Loewenschuss, A. J . Chem. Phys. 1976, 64, 1967. (38) Goggin, P. L.; Kemeny, G.; Mink, J. J . Chem. Soc., Faraday Trans. 2 1976, 72, 1025. (39) Roberts, H. L. Adv. Inorg. Chem. Radiochem. 1968, 1 1 , 309. (40) Lobana, T. S.; Manindejeet, K. S. J. Chem. Soc., Dalton Trans. 1988, 2913. (41) (a) Bell, N. A.; March, L. A.; Nowell, I. W. Inorg. Chim. Acta 1989, 156,201. (b) Bell, N. A.; March, L. A.; Nowell, I. W. Inorg. Chim. Acta 1989, 162, 57. (42) (a) Bell, N. A.; Dee, T. D.; Goldstein, M.; McKenna, P. J.; Nowell, I. W. Irwrg. Chim. Acta 1983, 71, 135. (b) Buergi, H. B.; Fischer, E.; Kunz, R. W.; Parvez, M.; hegosin, P. S. Inorg. Chem. 1982, 21, 1246. (43) Cecconi, F.; Ghilardi, C. A.; Midollini, S.;Orlandi, A. Inorg. Chim. Acta 1994, 271, 155. (44) (a) Reed, A. E.; Weinstock, R. B.; Weinhold, F. J . Chem. Phys. 1985, 83, 735. (b) Reed, A. E.; Weinhold, F. J . Chem. Phys. 1985, 83, 1736. (c) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88, 899.
JP950567F
