Zhang Lu-Lu, Song Yu-Zhi, Gao Shou-Bao, Zhang Yuan, Meng Qing-Tian. Accurate double many-body expansion potential energy surface of HS2A2A′) by scaling the external correlation. Chinese Physics B, 2016, 25(5): 053101
Permissions
Accurate double many-body expansion potential energy surface of HS2A2A′) by scaling the external correlation
Project supported by the National Natural Science Foundation of China (Grant No. 11304185), the Taishan Scholar Project of Shandong Province, China, the Shandong Provincial Natural Science Foundation, China (Grant No. ZR2014AM022), the Shandong Province Higher Educational Science and Technology Program, China (Grant No. J15LJ03), the China Postdoctoral Science Foundation (Grant No. 2014M561957), and the Post-doctoral Innovation Project of Shandong Province, China (Grant No. 201402013).
Abstract
Abstract
A globally accurate single-sheeted double many-body expansion potential energy surface is reported for the first excited state of HS2 by fitting the accurate ab initio energies, which are calculated at the multireference configuration interaction level with the aug-cc-pVQZ basis set. By using the double many-body expansion-scaled external correlation method, such calculated ab initio energies are then slightly corrected by scaling their dynamical correlation. A grid of 2767 ab initio energies is used in the least-square fitting procedure with the total root-mean square deviation being 1.406 kcal·mol−1. The topographical features of the HS2(A2A′) global potential energy surface are examined in detail. The attributes of the stationary points are presented and compared with the corresponding ab initio results as well as experimental and other theoretical data, showing good agreement. The resulting potential energy surface of HS2(A2A′) can be used as a building block for constructing the global potential energy surfaces of larger S/H molecular systems and recommended for dynamic studies on the title molecular system.
Sulfur-containing species have received growing attention due to their important role played in the air pollution and acid rain.[1–7] Among various sulfur-containing compounds, the HS2 radical is one of the most important species, which is responsible for the oxidation of reduced forms of sulfur, the reaction in combustion[8–10] and biochemistry.[11,12] Therefore, extensive studies on the HS2 radical have been carried out both theoretically and experimentally for decades.
Among a vast array of theoretical studies, Sannigrahi et al.[13] performed CI calculations and provided both structural and vibrational information of the electronic ground and excited states of HS2. Zhou et al.[14] carried out investigations on the geometries and spectroscopic parameters of the XS2 (X = H, F, Al, and Br) radicals using the split-valence plus polarization function and larger basis sets at the self-consistent field (SCF) and the second order Møller–Plesset (UMP2). By employing the CCSD(T) theory, the equilibrium structure and vibrational frequencies were reported by Owens et al.[15] for the X2H hydrides (X = Al, Si, P, and S) at their global minima. Subsequently, more accurate structural and thermodynamic properties of the HS2 radical were obtained by Denis[16] using the CCSD(T) and B3LYP levels of theory. Later, the accurate three-dimensional, near-equilibrium potential energy and dipole moment functions for both the ground and the first electronic excited states of HS2 were constructed by Peterson et al.[17] by using highly correlated coupled cluster methods with explicit basis set extrapolation. Song et al.[18] reported an accurate double many-body expansion (DMBE) potential energy surface (PES) for ground-state HS2 based on ab initio energies extrapolated to the complete basis set limit. Quite recently, Qin et al.[7] calculated the equilibrium geometry of the first excited electronic state of HS2 by time-dependent density functional theory using the hybrid functional B3LYP with the aug-cc-pVQZ (AVQZ) basis set.
The vibrational frequencies of the ground state (X2A″) and the excited state (A2A′) of HS2 were investigated by Holstein et al.[19] By studying the microwave spectroscopy of the ground electronic state of HS2, Yamanoto and Saito[20] reported the lowest energy structure. Ashworth et al.[21] first observed the far infrared laser magnetic resonance (LMR) spectrum of the ground state of HS2. Subsequently, Isoniemi et al.[22] investigated more refined structural information of HS2 radical in Ar matrix following the 266 nm photolysis of H2S2. The S–H stretching and the S–S–H bending modes were observed at 2463 cm−1 and 903 cm−1, respectively. Ashworth and Fink[23] analyzed the X–A band system through high-resolution Fourier-transform measurements of the chemiluminescence spectra of both HS2 and DS2. Entfellner and Boesl[24] presented the photodetachment-photoelectron spectra of the ground and the first excited states of HS2 and DS2 with a conventional time of flight photoelectron spectrometry.
Indeed, as far as we are aware, there is no work toward obtaining a global PES for the first electronic excited state of HS2(A2A′) reported, which would be of interest to the investigations of the H(2S) + S2(a1Δg)→ SH(X2Π) + S(3P) reaction. Thus, the major goal of the present work is to obtain a high quality global PES of HS2(A2A′) based on the DMBE[25–28] methodology.
This paper is organized as follows. Section 2 describes the ab initio calculations employed in the present work. Section 3 presents a brief description of the analytical DMBE formalism. Then the major topographical features of the HS2(A2A′) PES are discussed in Section 4. The concluding remarks are summarized in Section 5.
2. Ab initio calculations
The ab initio calculations have been carried out at the multi-reference configuration interaction level[29,30] of theory, including the Davidson correction [MRCI(Q)],[31] using the full valence complete-active-space self-consistent field (CASSCF)[32] wave function as the reference. The AVQZ basis set of Dunning[33,34] is employed, and the ab initio calculations are performed using the Molpro 2012 package.[35] A grid of 2767 ab initio points is chosen to map the PES over the H–S2 region defined by 2.5 ≤ RS2/a0 ≤ 5.5, 0.2 ≤ rH−S2/a0 ≤ 15, 0° ≤ γ/≤ 90°, and for the S–SH interactions, 2.0 ≤ RSH/a0 ≤ 4.0, 0.6 ≤ rS−SH/a0 ≤ 15, 0° ≤ γ/≤ 180°, where R, r, and γ are the atom–diatom Jacobi coordinates. The Cs point group symmetry is employed in the ab initio calculations, which holds two irreducible representations A′ and A″. For HS2(A2A′), seven A′ and two A″ symmetry molecular orbitals (MOs) are determined as the active space, amounting to 214 (136A′ + 78A″) configuration state functions. The MOs which correlate with the 3s and 3p atomic orbitals of each S and the 1s orbital of the H atom are utilized to determine the valence space.
The ab initio energies calculated in this way have been subsequently corrected using the double many-body expansion-scaled external correlation method (DMBE-SEC)[36] to account for the excitations beyond singles and doubles, most importantly, for incompleteness of the basis set. According to the DMBE-SEC scheme, the total DMBE-SEC interaction energy is written as
where
The first terms of Eqs. (2) and (3) sum over all the diatomic components with i = AB, BC, AC, and R = {RAB, RAC, RBC} in and is a collective variable of all internuclear distances. Then and in Eq. (3) are expressed as
Following the previous work,[37–39] the scaling factors in Eq. (4) are chosen to reproduce the corresponding bond dissociation energies of the i-th diatom, while F(3) in Eq. (5) is estimated as the average of the three two-body F-factors. The scaling factors obtained with the AVQZ basis set are , , and F(3) = 0.8134.
3. Analytic form of DMBE potential energy surface
In the framework of DMBE methodology, the single-sheeted PES can be written as
where R = (R1,R2,R3) is a collective diatomic internuclear separation, and the two-body and the three-body energy terms are split into two contributions: the extended Hartree–Fock (EHF) and the dynamical correlation (dc) energies. The following subsections give a detailed description of the two-body and the three-body energy terms in Eq. (6).
3.1. Two-body energy terms
The diatomic potential energy curves (PECs) of S2(a 1Δg) and SH(X2Π), which show the correct behaviors at the asymptotic limits R→ 0 and R → ∞, have been modeled using the EHF approximate correlation energy method.[40,41]
The EHF-type energy term is written as
where
Here r = R − Re is the displacement from the equilibrium diatomic geometry; D, ai (i = 1,…,7), and γi (i = 0,1,2) are adjustable parameters to be obtained from a least-square fitting procedure.
with the damping functions for the dispersion coefficients assuming the form
where An = α0n−α1 and Bn = β0exp(−β1n) are auxiliary functions. The α0, β0, α1, and β1 are universal dimensionless parameters for all isotropic interactions: α0 = 16.36606, α1 = 0.70172, β0 = 17.19388, and β1 = 0.09574. Moreover, ρ/a0 = 5.5 + 1.25R0, where is the LeRoy parameter,[43] and and are the expectation values of the squared radii for the outermost electrons in atoms A and B, respectively.
3.2. Three-body energy terms
As usual,[44,45] the three-body dc energy term is written in the following functional form:
where Ri is the corresponding diatomic internuclear separation, ri is the distance between atom A and the center-of-mass of diatom BC, and θi is the angle between these two vectors. fi(R) = {1 − tanh[ξ (ηRi − Rj − Rk)]}/2 is a switching function. As was done in Ref. [37], we have used η = 6 and ξ = 1.0a0−1. The χn(ri) in Eq. (11) is a damping function, which is written in the functional form employed elsewhere.[37,44] As was done for NH2(22A″),[46] the parameters of are taken from Ref. [18] for the ground electronic state of HS2.
For a given triatomic geometry, by removing the sum of the two-body energy terms from the corresponding interaction energies, one obtains the total three-body energy. Then by subtracting the three-body dc contribution from the total three-body energy, the three-body EHF energy contribution is obtained. It can be modeled in the three-body distributed-polynomial[47] form
where Pj(Q1,Q2,Q3) is the j-th polynomial up to six-order in symmetry coordinates, which is defined as
In the above equation, , , and , with Qi (i = 1,2,3) being the symmetry coordinates defined as
In turn, are the reference geometries and are the nonlinear range-determining parameters, which are optimized via a trial-and-error least-squares fitting procedure[48,49] that minimizes the root-mean-squared deviation (rmsd) error while warranting the proper behavior in the asymptotic region. The complete set of parameters totals 150 linear coefficients, 9 nonlinear ones, and 9 reference geometries. Table 1 shows the rmsd values of the final DMBE/SEC PES with respect to all the fitted ab initio energies. As shown in Table 1, a total of 2676 points have been used for the calibration procedure, thus covering a range up to 400 kcl·mol−1 above the HS2 global minimum.
Table 1.
Table 1.
Table 1.
Root-mean-square deviations (in kcal·mol−1) of the PES.
.
Energy/kcal·mol−1
Na
rmsd
10
24
0.206
12
20
65
0.491
14
30
133
0.925
34
40
217
1.046
54
50
321
1.103
91
60
769
1.221
148
70
1157
1.279
193
80
1286
1.321
224
90
1703
1.276
307
100
1807
1.289
333
200
2520
1.341
443
300
2631
1.371
463
400
2676
1.406
46
Number of points in the indicated energy range.
Number of points with an energy deviation larger than the rmsd.
Table 1.
Root-mean-square deviations (in kcal·mol−1) of the PES.
.
4. Features of potential energy surface
The fitted potential energy curves of S2(a 1Δg) and SH(X2Π) and the deviations from the MRCI(Q)/AVQZ energies are displayed in Fig. 1. As shown in Fig. 1, the modeled PECs show smooth behaviors and accurately mimic the ab initio energies, with the maximum error being smaller than 0.05 kcal · mol−1. Parameters gathered in Table 2 include the equilibrium geometry (Re), dissociation energy (De), harmonic vibrational frequency (ωe), non-harmonic vibrational frequency (ωexe), and rotational constants (αe and βe). Excellent agreement can be found in Table 2 between the results obtained from the modeled PECs and the experimental data. For SH(X2Π), the deviations for Re, De, ωe, ωexe, αe, and βe are calculated to be 0.0002 a0, 0.3948kcal · mol−1, 5.736 cm−1, 0.447 cm−1, 0.0346 cm−1, and 0.1304 cm−1, respectively, while the corresponding deviations for S2(a 1Δg) are 0.0199 a0, 0.15 cm−1, 0.845 cm−1, 0.00017 cm−1, and 0.0033 cm−1. Good agreement is also found compared to other theoretical results.[17,18,50–53]
Fig. 1. Potential energy curves of (a) S2(a 1Δg) and (b) SH(X2Π), and the differences between the fitted PECs and the ab initio points.
Table 2.
Table 2.
Table 2.
Spectroscopic constants of SH and S2 diatoms, Re is in the units of a.u., De is in the units of kcal · mol−1, and ωe, ωexe, αe, and βe are in the units of cm−1.
The dissociation energies are calculated by De = D0 + ωe/2.
Obtained at CCSD(T)/AV5Z level of theory.
Table 2.
Spectroscopic constants of SH and S2 diatoms, Re is in the units of a.u., De is in the units of kcal · mol−1, and ωe, ωexe, αe, and βe are in the units of cm−1.
.
Figures 2–4 illustrate the major topographical features of the HS2(A2A′) DMBE/SEC PES, and the corresponding structural parameters of the stationary points are collected in Table 3, including our ab initio values calculated at the MRCI(Q)/AVQZ level of theory and other theoretical and experimental results. The global minimum obtained from the present DMBE/SEC PES locates at R1 = 3.937a0, R2 = 2.540a0, and R3 = 4.836a0 (R1 is the SS interatomic separation, while R2 and R3 are the two SH interatomic separations), differing only 0.025a0, 0.002a0, and 0.003a0 from the MRCI(Q)/AVQZ results, respectively. Compared with the theoretical works of Peterson et al.[17] and Denis,[16] the maximal differences occur in R3, which are only 0.043 a0 and 0.025 a0, respectively. Since the DMBE-SEC method is employed to model the ab initio energies, the well depth of the global minimum is 0.0078Eh lower than that of MRCI(Q)/AVQZ. As for the harmonic frequencies, the DMBE/SEC PES from the present work predicts values of 2714.19 cm−1, 504.45 cm−1, and 812.27 cm−1, which are in good agreement with the most recent experimental values given by Qin et al.,[7] which are 2585 ± 81 cm−1, 490 ± 41 cm−1, and 741 ± 41 cm−1, respectively. Table 3 also gathers the properties of three saddle points (SP): SP1 (D∞v), SP2 (C∞v), and SP3 (C∞v), and two transition states (TS): TS1 (C2v) and TS2 (C∞v).
Fig. 2. Contour plots of HS2 (A2A′) potential energies for the four H–S–S angles (a) 0°, (b) 45°, (c) 90°, and (d) 135°. Contours are equally spaced by 0.006Eh, starting from −0.1668Eh in panel (a), −0.1977Eh inpanel (b), −0.2294Eh in panel (c), and −0.2006Eh in panel (d).
Fig. 4. The spherically averaged isotropic (V0) and leading anisotropic potentials (V2): (a) for H + S2 interaction, with RSS = 3.6079a0; (b) for S + SH interaction, with RSH = 2.5337a0.
Table 3.
Table 3.
Table 3.
Properties of stationary points on the fitted HS2(A2A′) PES (harmonic frequencies in cm−1).
Properties of stationary points on the fitted HS2(A2A′) PES (harmonic frequencies in cm−1).
.
The contour maps for the H(2S) + S2(a1Δg) → SH(X 2Π) + S(3P) reaction in internal coordinates at four H–S–S angles (∠HSS = 0°, 45°, 90°, and 135°) are shown in Fig. 2. The contours are equally spaced by 0.006Eh, starting at −0.16681Eh for 0°, −0.19773Eh for 45°, −0.22936Eh for 90°, and −0.20061Eh for 135°. The notable feature observed is that there are two valleys in each map: the left upper valley corresponds to H(2S) + S2(a1Δg) asymptote, while the valley at the right bottom corresponds to SH(X2Π) + S(3P) asymptote. It can also be seen in Fig. 2 that deep wells connect the two asymptotes. For a better understanding of the characteristics of the present PES, the minimum energy paths (MEPs) of the H(2S) + S2(a1Δg) → SH(X 2Π) + S(3P) reaction at the four H–S–S angles are shown in Fig. 3. These MEPs represent the potential energies of HS2(A2A′) as a function of a suitable reaction coordinate defined as RS2 − RSH, where RS2 and RSH are the S–S and S–H internuclear distances, respectively. The RS2 approaches the S2 equilibrium distance when the reaction coordinate is large, whereas a large positive reaction coordinate corresponds to RSH approaching the SH equilibrium distance. The exothermicity of the H(2S) + S2(a1Δg)→SH(X2Π) + S(3P) reaction is 3.6 kcal · mol−1. The salient feature when ∠HSS = 0°, 45°, 90°, and 135° is that there exist the deep wells at 17.30 kcal · mol−1, 36.85 kcal · mol−1, 56.45 kcal · mol−1, and 38.56 kcal · mol−1 relative to the H + S2 reactant, respectively. As shown in Fig. 3, there also exist barriers when ∠HSS = 0°, 45°, 90°, and the highest barrier (6.459 kcal · mol−1 relative to H + S2 reactant) is obtained when ∠HSS = 0°, locating at RS2 = 3.652a0 and RSH = 4.265a0.
Figure 4 shows the spherically averaged isotropic (V0) and leading anisotropic potentials (V2) for H + S2 and S + SH scattering processes. The VL (L = 0,2) potentials are obtained by expressing the interaction energies in a Legendre expansion[54–56]
where Re are the internuclear separations of S2 and SH, which are fixed at their equilibrium values of 3.6079a0 and 2.5337a0, r and θ are the Jacobi coordinates, and PL are the Legendre polynomials. Then, the VL potentials can be calculated by the following integral:[57]
The isotropic average potential (V0) determines how close on average the atom and the diatom can approach each other. In turn, the sign of V2 indicates whether or not the molecule prefers to orient its axis along the direction of the incoming atom: a negative value favors the collinear approach, while a positive value favors the approach through an isosceles triangular geometry. As shown in Fig. 4(a), V2 is positive at distances larger than 5.50a0 and smaller than 8.20a0, which indicates that the H atom favors approaching the S2 diatom in this region through an isosceles triangular geometry. For the S + SH interaction displayed in Fig. (b), there is no barrier in V0, while the V2 component is non-negative in the regions of interest, which corroborate the fact that the interaction favors the insertion of the S atom perpendicularly to the SH diatom.
5. Conclusion
A global accurate PES is reported for HS2(A2A′) by fitting the MRCI(Q)/AVQZ energies which are scaled by the DMBE-SEC method. The resulting PES is realistic over the entire configuration space. The various topographical features of the current DMBE/SEC PES have been examined in detail and compared with the previous theoretical and experimental results. The properties, including geometries, energies, and vibrational frequencies, have been characterized on the current HS2(A2A′) DMEB-SEC PES. The results show good agreement with the previous high quality theoretical studies and the available experiment data. Thus, the PES constructed here can be used for dynamic studies of reactions and also as a building block for the PESs of larger molecular systems containing the HS2 fragment.