ebook img

Experimental, semi-experimental and ab initio equilibrium structures PDF

154 Pages·2017·4.04 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Experimental, semi-experimental and ab initio equilibrium structures

Molecular Physics F o r P e Experimental, semi-experimental and ab initio equilibrium estructures r Journal: Molecular Physics R Manuscript ID: TMPH-2007-0302 e Manuscript Type: Invited Article v Date Submitted by the i 11-Oct-2007 Author: e Complete List of Authors: Demaison, Jean; Université de Lille I, PhLAM Keywords: equilbrium structure, ab initwio, spectroscopy O n l y URL: http://mc.manuscriptcentral.com/tandf/tmph Page 1 of 152 Molecular Physics 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 F 17 18 o 19 20 Experimentarl, semi-experimental and ab initio equilibrium structures 21 22 23 P J. Demaison 24 25 e 26 e 27 28 Laboratoire de Physique des Laserrs, Atomes, et Molécules, UMR CNRS 8523, Université de 29 30 Lille I, F-59655 Villeneuve d'Ascq Cédex, France 31 R 32 33 e 34 E-mail: [email protected] ; fax: +33 3 20 33 70 20 35 v 36 i 37 38 e 39 40 w 41 42 43 O 44 45 46 n 47 l 48 y 49 50 51 52 53 Number of pages: @ 54 55 Number of tables: 21 56 57 Number of figures: 4 58 59 60 1 URL: http://mc.manuscriptcentral.com/tandf/tmph Molecular Physics Page 2 of 152 1 2 3 Abstract 4 5 The determination of equilibrium structures of molecules by spectroscopic methods or 6 7 by quantum mechanical calculations is reviewed. The following structures are described in 8 detail: experimental equilibrium structures, empirical structures, semi-experimental structures 9 10 and ab initio structures. The approximations made by the different methods are discussed and 11 12 their accuracies are compared. 13 14 15 16 F 17 1. Introduction 18 o 19 The concept of molecular structure was developed during the 19th century. Couper [1], 20 r 21 around 1856, was one of the first, with Kekulé [2], to state that carbon atom is tetravalent and 22 23 to form a concrete idea Pof molecular structure. In 1885, von Baeyer [3] published his theory 24 of ring strain whose main conclusion is that bond angles may deviate from the tetrahedral 25 e 26 value of 109°28'. In 1874, the concept of tetrahedral carbon was simulteanously introduced by e 27 28 le Bel [4]and van't Hoff [5]. r 29 30 The golden age of molecular structure research is the first half of the 20th century, 31 R perhaps starting with the discovery of X-ray diffraction around 1900 permitting to determine 32 33 interatomic distances in crystals [6]. The setructure of diamond was obtained in 1913 by the 34 35 Bragg's, father and son [7]. Debye [8] was thev first to determine the experimental structure of 36 i 37 a molecule (CCl ) in the gas phase by X-ray diffraction. At about the same time, around 1930, 4 38 e the first gas-phase electron diffraction experiment was performed by Mark and Wierl [9]. 39 40 w Although the accuracy of the measurements was not high, many molecules were studied in a 41 42 relatively short time and the results were used to develop concepts concerning chemical 43 O 44 bonding by Pauling [10] and others. 45 46 In quantum chemistry, the first successful attempt to ncalculate the structure of a 47 l molecule was by Burrau in 1927 on H + [11]. It was followed, in the same year, by the 48 2 y 49 calculation of the bond length in H by Heitler and London [12]. 50 2 51 After the second world war microwave spectroscopy had arrived [13] which is 52 53 generally considered to be the most precise technique for obtaining molecular geometries in 54 the gas phase. This spectroscopy (as well as high resolution infrared spectroscopy) now 55 56 determines rotational constants with a precision close to 1 in 108. If there was a simple 57 58 relation between experimental rotational constants and equilibrium geometry, it would be 59 60 possible to determine the structure of molecules with a tremendous accuracy. However, we will see that many factors limit this accuracy. 2 URL: http://mc.manuscriptcentral.com/tandf/tmph Page 3 of 152 Molecular Physics 1 2 3 The apparition of the computer and its considerable increase in power are also 4 5 extremely important because it permitted the development of ab initio electronic structure 6 7 methods which can now provide accurate estimates of equilibrium structures. 8 There are many reviews devoted to structure determination. However, most of the 9 10 time, they limit themselves either to a single experimental technique or to ab initio 11 12 calculations. There are relatively few papers where a critical comparison of the experimental 13 14 and ab initio techniques is made and where the interplay of these methods is emphasized. 15 Furthermore, although the techniques are described in great detail, there is no thorough 16 F 17 discussion of the accuracy really achievable. 18 o 19 The first question to answer is which accuracy is desirable. There are three reasons 20 r 21 militing in favour of a high accuracy: 22 23 • Theoreticians nePed accurate structures to check their more and more sophisticated 24 computations. Although the accuracy of ab initio calculations varies wildly, it is 25 e 26 reasonable to define a range from 0.3 pm to better than 0.1 pm. e 27 28 • Inspection of the range of ra few X-Y bond lengths shows that it is rather small. For 29 30 instance, it is only 6 pm for the CH and NH bonds in different molecules. Thus, to 31 R compare a particular structure in different molecules, an accuracy significantly better 32 33 than 1 pm is required. e 34 35 • The energy of a molecule is sensitive vto its structure. Molecular mechanics programs 36 i 37 for the calculation of properties of large molecules are parametrized against a small 38 e number of small molecules whose structure is assumed to be accurate. For instance, 39 40 distortion of a C-C single bond by 2 pm "costws" 0.6 kJ mol-1; distortion of a ∠(CCC) 41 42 bond angle by 2°, about 0.4 kJ mol-1; while the torsional distortion of a CCCC chain 43 44 by 5° costs about 0.2 kJ mol-1 [14]. In order to be ablOe to determine the relative energy 45 46 of a molecule with an acceptable accuracy (a few kJ/moln), it is necessary to scale the 47 l molecular mechanics programs [15] with the aid of molecules whose molecular 48 y 49 geometry is very accurately known. 50 51 The next step is to see under which conditions such an accuracy can be achieved. For this 52 53 goal, we will first analyze the approximations which are made during the derivation of the 54 55 molecular Hamiltonian. 56 57 58 59 60 2. The molecular Hamiltonian [16] 3 URL: http://mc.manuscriptcentral.com/tandf/tmph Molecular Physics Page 4 of 152 1 2 3 A molecule is a collection of nuclei and electrons, held together by certain forces and 4 5 obeying the laws of quantum mechanics. The dominant forces are the Coulomb electrostatic 6 7 forces. Gravitational forces are also present but a simple calculation shows that they are 8 completely negligible compared to the electrostatic forces. For open-shell molecules or when 9 10 heavy nuclei are present, the electron spin magnetic moment may interact with other magnetic 11 12 moments (generated by the orbital motion of the electrons or of the nuclei, or magnetic 13 14 moments of the other electrons). These interactions are much smaller and will be neglected in 15 order to simplify the presentation. When they are not negligible, they can be taken into 16 F 17 account without difficulty1. What is left is the Coulomb Hamiltonian. To write it, it is 18 o 19 assumed that the nuclei are point masses and the relativistic effects are negligible. The first 20 r 21 approximation is a ve ry good one and will be further discussed in section 5.2. The second 22 23 approximation is also a Psound one. However, when heavy nuclei are present or during highly 24 accurate electronic structure computations, relativistic effects become important and should 25 e 26 be taken into account. This approximation will be further discussed in section 3.5. Using these e 27 28 approximations, the molecular Hamriltonian may be written as 29 30 31 R 32 H =TN+(Te+VNe+VNN+Vee)=TN+He (1) 33 e N 3345 with TN =−h2∑2ΔMα v (2a) 36 € α=1 α i 3378 T =−h2∑n Δi e (2b) e m 2 39 € i=1 40 w N n 41 1 Z V =− ∑∑ α (2c) 42 Ne 4πε r 43 € 0α=1i=1 iα O 44 1 N n Z Z 45 V = ∑∑ α β (2d) NN 4πε R 46 € 0α=1β>α αβ n 47 l 48 1 n n 1 V = ∑∑ y (2e) 49 ee 4πε r 50 € 0 i=1 j>i ij 51 ∆ is the Laplace operator, M and Z are the mass and atomic number of the nucleus α, m is 52 α α 5534 € the mass of the electron, riα is the distance between electron i and nucleus α, and similar 55 definitions hold for r and R . ij αβ 56 57 58 59 60 1 The interactions of the magnetic and electric moments of the nuclei with the other electric and magnetic moments in the molecule may lead to a splitting of the energy levels in several hyperfine components but they do not affect the following discussion. 4 URL: http://mc.manuscriptcentral.com/tandf/tmph Page 5 of 152 Molecular Physics 1 2 3 2.1. The Born-Oppenheimer approximation and the electronic Hamiltonian [17] 4 5 This Hamiltonian is too complicated to be solved exactly. As proposed by Born and 6 7 Oppenheimer (BO), the nuclear kinetic energy T is first separated when motion of the N 8 electrons is investigated. The justification of the BO approximation is that the heavy nuclei 9 10 move much more slowly than the light electons (it also assumes that the momentum of the 11 12 electrons and the nuclei is of the same order of magnitude), thus T is an almost constant term N 13 14 and can therefore be neglected in the differential equation. In the remaining electronic 15 Hamiltonian H , the nuclear positions R enter as parameters 16 e F 17 H ψ(e) = EBOψ(e) (3) 18 e o 19 Varying the position of R in small steps, one obtains EBO as a function of R. This is the 20 r 21 potential energy (hype r)surface (PES) : EBO(R). Its minimum corresponds to the equilibrium 22 23 structure of the moleculeP. As the nuclear masses are absent in He, EBO is isotopically invariant. 24 25 In other words, in this approeximation all the isotopologues of an individual molecule have the 26 same PES and the same equilibrium structure irrespective of the coordinate system used to e 27 28 represent R. It can be shown thrat the BO approximation can be trusted when the PESs 29 30 corresponding to the different electronic states are well separated: 31 R 32 E 0BO<<E1BO<<E2BO<<L (4) 33 e This is generally a good approximation for ground electronic states except for a few 34 35 v molecules. For instance, the first two excited electronic states of NO are close to the ground 36 € i 3 37 state and all three levels interact via vibronic couplings [18]. 38 e 39 The PES and the equilibrium structure can be calculated at different levels of electronic 40 w 41 structure theory. This will be discussed in sections 3 and 6 together with the relativistic effects 42 and the correction to the BO approximation. 43 O 44 45 46 2.2. Vibration-rotation Hamiltonian [19] n 47 l 48 In the second step of the BO approximation the nuclear kinetic energy T is N y 49 reintroduced and the Hamiltonian is written as 50 51 [T + EBO(R)]φ(R) (5) 52 N 53 where EBO(R) may be considered as the potential. 54 55 The kinetic part of the vibration-rotation Hamiltonian was first derived by Wilson and 56 57 Howard [20]. However, this original Hamiltonian is rather complicated (particularly due to 58 the non-commutation of the various terms). Watson [21] could considerably simplify it. This 59 60 led him to introduce a small correction term U which acts as a small mass-dependent effective 5 URL: http://mc.manuscriptcentral.com/tandf/tmph Molecular Physics Page 6 of 152 1 2 3 contribution to the potential energy. It comprises the translation of the center of mass 4 5 described by three degrees of freedom (dof), the overall rotation of the molecule (three dofs) 6 7 and the nuclear vibrations (3n – 6 dofs). The separation of the translational motion is 8 straightforward but the (approximate) separation of vibrational and rotational motion is not a 9 10 trivial problem. It is treated in many textbooks [16, 19, 20]. 11 12 After separation of rotation and vibration, a rotational Hamiltonian may be written for each 13 14 vibrational state v as 15 16 Hv = ∑BξJ2 +H , (6) R Fv ξ cd 17 ξ=a,b,c 18 19 where B ξv is the rootational constant in vibrational state v about the principal axis ξ, J ξ2 is the 20 r 21 € component of the rotat ional angular momentum along the principal axis ξ and Hcd is a small 22 23 €c orrection term called cPentrifugal distortion which is due to the fact that the€ molecule is not 24 rigid and thus, upon rotation, the molecule is distorted by the stretching effects of the 25 e 26 centrifugal forces. e 27 28 In a perturbational treatment, the rortational constant Bξ is given by v 29 30  d   d  d  31 Bξv =Beξ −∑αξi  vi+ 2i +∑γξij vi+ 2Ri vj+ 2ji +L (7) 32 i i≥j € 33 e The summations are over all vibrational states, each characterized by a quantum number v 34 i 3356 € and a degeneracy di. The parameters α ξi anvd i γ ξij are called vibration-rotation interaction 37 constants of different order. Bξ is the equilibrium rotational constant. It is proportional to the 38 e e 39 inverse of the equilibrium mom€en t of in€er tia which is itself a function of the coordinates at 40 w 41 equilibrium, € 42 43 h2 Bξ = O (8) 44 e 8π2Iξ(r ) 45 e e 46 The convergence of the series expansion is usually fast, αξ nbeing about two orders of 47 i l 48 € magnitude smaller than Bξ and γξ two orders of magnitude smaller than αξ, see Table 1. 49 e ij y i 50 The original rotational Hamiltonian cannot b€e used to analyze rovibrational spectra 51 52 because it conta€i ns ma€n y terms which are not all independent€. Watson [22] submitted the 53 54 Hamiltonian to unitary transformations which allowed him to reduce the number of 55 parameters. The rotational constants are only marginally affected by the transformation. The 56 57 rotational constants of an asymmetric top obtained from a fit using the Hamiltonian of Watson 58 59 are affected by a small centrifugal distortion contribution which depends on the choice of the 60 6 URL: http://mc.manuscriptcentral.com/tandf/tmph Page 7 of 152 Molecular Physics 1 2 3 reduction and of the representation. Watson has shown that only the following linear 4 5 combinations can be determined from the analysis of the spectra: 6 7 8 B = B(A) +2Δ = B(S) +2D +6d (9a) 9 z z J z J 2 10 B = B(A) +2Δ + Δ −2δ −2δ = B(S) +2D + D +2d +4d (9b) 11 x x J JK J K x J JK 1 2 12 B = B(A) +2Δ + Δ +2δ +2δ = B(S) +2D + D −2d + 4d . (9c) 13 y y J JK J K y J JK 1 2 14 In eqs.(9) B(A) are the experimental constants in the so-called A reduction, B(S) are the 15 ξ ξ 16 17 experimentalF constants in the so-called S reduction, and B are the determinable constants ξ 18 19 (wh€e re ξ = x, y, oz). However, these latter constants are still contaminate€d by the centrifugal 20 r 21 distortion. As shown b y Kivelson and Wilson [23], the rigid rotor constants B ' are given by ξ 22 1 1 23 B′ =B + (τ +τP+τ )+ τ x x 2 yyzz xyxy xzxz 4 yzyz (10) 24 25 e € 26 where B′ and B′ are obtained by cyclic permutation of x, y, and z. y z e 27 € 28 Finally, Watson has shown that thre rotational constants are affected by the mass-dependent 29 30 contribution U to the potential energy. This contribution may be written as 31 1 R 32 Brigid =B′ − [τ +τ +τ ] (11) 33 x x 8 xxxx xxyy xxzz e 34 The problem is that the τ constants are experimentally determinable only for a planar 35 v 36 € molecule by means of the planarity relations of Diowling [24]. For a non-planar molecule they 37 38 can be calculated from the harmonic force field [25e]. 39 These equations are approximate for three reasons: i) the result is sensitive to the particular 40 w 41 choice of the rotational constants (equilibrium, ground state experimental, ground state 42 43 corrected, etc.) used in the calculations ii) the higher-order terms (sextic terms) are neglected O 44 45 iii) there are still additional terms as shown by Chung and Parker [26] but whose expressions 46 n 47 are not known to date. However, as noted by Aliev and Watson [27]l these last corrections are 48 indistinguishable from the effects of the breakdown of the BO approxiymations. 49 50 Compared to the other corrections, the centrifugal distortion correction is generally 51 52 quite small except for very light molecules, see Table 2. Furthermore, it is different from zero 53 54 only for asymmetric top molecules. However, in this case, it generally remains much larger 55 than the experimental accuracy. In many practical cases, the spectroscopist will usually fit the 56 57 line frequencies measured in order to obtain what he takes to be the experimental ground state 58 59 rotational constants plus all quartic (for sufficiently high J also the sextic) centrifugal 60 distortion constants. He then uses these rotational constants to determine the molecular 7 URL: http://mc.manuscriptcentral.com/tandf/tmph Molecular Physics Page 8 of 152 1 2 3 structure. It should be better to use the determinable constants, or still better to use the ground 4 5 state rigid rotor corrected constants of Eq. (11). The spectroscopist seldom goes farther than 6 7 to the determinable constants. 8 9 10 2.3. Magnetic correction [28] 11 12 Since the electrons tend to follow the motion of the nuclei, the bulk of the electronic 13 14 contribution to the rotational constants can be taken into account by employing atomic rather 15 than nuclear masses. This is a very good approximation for most molecules and it is about the 16 F 17 only feasible one for polyatomic molecules. However, a small correction for unequal sharing 18 o 19 of the electrons by the atoms and for nonspherical distribution of the electronic clouds around 20 r 21 the atoms is sometimes nonnegligible and has to be taken into account. 22 23 The total angular momePntum J of a molecule may be written as the sum of N, the angular 24 momentum due to the rotation of the nuclei and L, the angular momentum of the electrons. 25 e 26 The rotational Hamiltonian for the nuclear system plus the Hamiltonian for the unperturbed e 27 28 electronic energies may be written ras 29 333012 H= 12∑ξ NIξξ2 +He = 12∑ξ (Jξ −IξLξ)2 +RHe 33 1 J2 J L 1 L2 e (12) 34 = ∑ ξ +H −∑ ξ ξ + ∑ ξ 35 2 I e I 2 I v ξ ξ ξ ξ ξ ξ 36 14 4 2 4 4 3 14 2 4 3 i 37 H0=HR+He H ′ 38 Since L is very small, the third term can be negleceted and H' can be treated as a perturbation 39 ξ 40 € of H0. We now assume that the molecule is not inw a pure 1Σ state ψ(0) (L = 0) but in a 41 0 42 perturbed state ψ(1) which has some electronic momentum. The correct effective rotational 0 43 O 44 Hamiltonian is then € 45 46 H€ = ψ(1) H +H′ ψ(1) n (13) 47 eff 0 R 0 l 48 A simple perturbation calculation up to second-order gives y 49 50 €  2 555123 Heff = 12∑ξ Jξ2 I1ξ − I2ξ2 ∑n≠0 nEnL−ξE00  (14) 54 55 This is equivalent to the definition of an effective moment of inertia (Iξ)eff by 56 57 € nL 0 2 58 1 = 1 − 2 ∑ ξ (15) 59 (Iξ)eff Iξ Iξ2 n≠0 En−E0 60 € 8 URL: http://mc.manuscriptcentral.com/tandf/tmph Page 9 of 152 Molecular Physics 1 2 3 where I on the right is calculated using the nuclear masses. This effective moment of inertia 4 ξ 5 can be expressed as a function of the molecular rotational g factor in the principal axis system, 6 7 whose definition is 8 2 910 g = Mp ∑Z (y2+z2)−2Mp ∑ nLx 0 (16) 11 xx Ix i i i i mIx n≠0 En−E0 12 and g and g are obtained by cyclic permutation. In this equation M is the mass of the 13 yy zz p 14 proton. 15 € 16 The effective rotational constant B (obtained from the analysis of the rotational spectrum) is F eff 17 18 therefore o 19 20 (Bξ) =Bξ + mrg Bn, (17) 21 eff M ξξ p 22 23 where Bξ is the rotatioPnal constant calculated with atomic masses and Bn the rotational 24 25 € constant calculated with nucelear masses. 26 The g factor can be obtained eexperimentally from the analysis of the Zeeman effect on the 27 28 r rotational spectrum [28, 29]. It can also be calculated ab initio [30]. A few typical results are 29 30 given in Table 3. As expected, the correction is the largest for very light molecules (as LiH) 31 R 32 and it rapidly decreases when the mass of the molecule increases. There are, however, a few 33 e exceptions. As the expression of g shows, see Eq. (16), g may become large when an 34 35 v electronic excited state is close to the ground state, (because the denominator E – E is 36 i n 0 37 small). This is the case for ozone (O ), where the magnetic correction is extremely large. 38 3 e 39 40 w 41 42 43 O 44 45 46 n 47 l 48 y 49 50 51 52 53 54 55 56 57 58 59 60 9 URL: http://mc.manuscriptcentral.com/tandf/tmph

Description:
n Lx 0. 2. En − E0 n≠0. ∑. (16) and gyy and gzz are obtained by cyclic permutation. In this equation Mp is the mass of the proton. The effective .. test set. This is true when the nondynamical correlation is small. When it is large, the situation is completely different as can be seen on the
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.