International Symposium on Molecular Spectroscopy 61st Meeting - - June 19-23, 2006 An Accurate ab initio Potential Energy Surface and Calculated Spectroscopic Constants for BeH2, BeD2 and BeHD υ Hui Li, Robert J. Le Roy Background • Chemical bond 2 2 Be(1s 2s ) sp hybridization Test new ab initio methods • Formation and Stability Be(g)+H2(g) 59.65 kcal/mol BeH2(g) (ΔE=-38.35 kcal/mol) • Solid BeH2 Linear(BeH2)n in matrix G. J. Brendel et al. Inorg. Chem. 17, 3589(1978) X. F. Wang et al. Inorg. Chem. 44, 610(2005) crystal structure G. S. Smith et al. Solid State Communications 67, 491(1988) Recent New Experiment • BeH2 detected in argon matrix T.J.Tague,Jr. et.al, J. Am. Chem. Soc. 115, 12111 (1993) • Free gaseous BeH2 Infrared emission spectrum of BeH2 and BeD2 P. F. Bernath, et al., Science, 297, 1323 (2002), A. Shayesteh, et al., J. Chem. Phys., 118, 3622 (2003) Previous Theoretical Results Theory Mode υ1 BeH2 BeD2 a Experiment f CC / CCSD(T)/ In Argon b a c [5s3p2d1f] cc-pCVTZ Matrix 1979.6 1987.0 υ2 716.8 705.2 697.9 υ3 2167.2 2157.8 2159.1 υ1 1420.8 υ2 545.1 531.9 υ3 1676.2 1674.0 Gaseous d BeH2 2178.9 1689.7 J. M. L. Martin and Timothy J. Lee, Chem. Phys. Lett. 200, 502 (1992). b T. Hrenar, H.-J. Werner & G. Rauhut, Phys. Chem. Chem. Phys. 7 3123 (2005). c T. J. Tague, Jr., and L. Andrews. J. Am. Chem. Soc. 115, 12111 (1993). d P. F. Bernath, et al., Science, 297, 1323 (2002) f A multi-level ab initio Scheme (1D: CCSD(T); 2D: MP4(SDQ), 3D: MP2) & VCI used Potential Energy Surface • • E(AV5Z): calculated at the icMRCI+Q/aug-cc-pV5Z level. Δcore: determined at MR-ACPF/cc-pCV5Z level E(cav5z)=E(av5z)+Δcore Transition state r(HH)=2.0250 a0 R= 2.6342 a0 θ=90º H Be H BeH2 Method Re(a0) a CMRCI/VQZ+K 2.506 2.5147 2.5067 2.5065 AV5Z AV5Z+Vcore b EXP. a Linear minimum 2.6342 a0 Re=2.5067 a0 Be+H2 TS. (eV) Be+H2 (eV) BeH+H (eV) H+Be+H (eV) 4.6633 4.2343 4.2493 J. Hinze et al., Mol. Phys. 96, 711 (1999) b 1.6208 1.6285 1.6628 4.2112 4.2247 6.3797 6.4231 P. F. Bernath, et al., Science, 297, 1323 (2002) Question: How is the accuracy of calculated vibrational levels affected by method for interpolating over the grid of PES Test 1: One by one, omit each grid point, interpolate for it from the rest, and determine error in the value obtained by different interpolation methods Test 2: Compare the vibrational levels using different interpolation methods RMSD (in -1 cm ) on spline interpolation over n V×Ri Range No. points n=0 n=2 n=4 All energies 6864 97.93 5.63 73.53 1857 6.60 0.89 3.53 6864 0.08 reference 0.04 Energies≤20 000 -1 cm 62 lowest pure levels of BeH2 Estimated the RMSD with n=2 interpolation is about 0.01 -1 cm Ro-vibrational Energy Levels • The Hamiltonian in Radau coordinates (R1, R2, ) is ˆ 1 1 1 1 1 J ˆ H ( )( sin ) 2 2 2 2 2m1 R1 2m2 R2 2m1 R1 2m2 R2 sin sin 2 2 2 z 2 ˆJ 2 Jˆ cot ˆ 1 ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ [( J i J ) ( J i J )] J [( J i J ) ( J i J )] V ( R , R , ) x y x y z x y x y 1 2 2 2 2 2m1 R1 2m1 R1 2m1 R1 2 2 z R1 and R2 are the radial Radau coordinates m1 and m2 are the masses of the two H atoms Jx, Jy, Jz are the components of the total angular momentum J The z axis of the body-fixed frame lies along the R1 radial Radau vector • A direct product discrete variable representation (DVR) • Lanczos recursion used to diagonalize the Hamiltonian Ro-vibrational Energy Levels and Wave Function (v1,v2,v3) (0,0,0) (0,0,1) (0,0,2) (1,0,0) (1,0,1) J=0 0.00 2178.79 4323.80 1991.97 4112.25 J=1 9.40 2188.05 4332.93 2001.26 4121.40 J=2 28.21 2206.58 4351.19 2019.84 4139.70 J=3 56.41 2234.37 4378.57 2047.71 4167.14 J=4 94.00 2271.41 4415.07 2084.84 4203.72 J=5 140.97 2317.70 4460.68 2131.25 4249.42 J=6 197.31 2373.21 4515.38 2186.91 4304.24 J=7 263.00 2437.93 4579.16 2251.80 4368.16 J=8 338.02 2511.85 4652.00 2325.92 4441.16 rs = R1,(BeH) + R2,(BeH) rd = R1, (BeH) - R2,(BeH) Spectroscopic Constant • The energy level expression for linear triatomic E G (1 , ,3 ) B[ J ( J 1) ] D[ J ( J 1) ] H [ J ( J 1) ] 2 0 2 2 2 2 3 1 2 3 [qJ ( J 1) qD [ J ( J 1)] qH [ J ( J 1)] ,1 2 G: pure vibrational energy level B: inertial rotational constant ℓ: vibrational angular momentum quantum D: centrifugal distortion constant ±: the parity forπ(ℓ=1) levels Σ state: q=qD=qH=0 • ℓ-type rotational resonances E H 2 W 20 0 E E E E 2W20 0 1 2 2 2 2 1/ 2 E W20 [q qD J ( J 1) qH J ( J 1) ( J ( J 1) 2 J ( J 1)) . 2 0 0 0 0 0 2 G. Amat et al. J. Mol. Spectrosc. 2, 163 (1958) A.G. Maki Jr. et al. J. Chem. Phys. 47, 3206 (1967) A. Shayesteh and P. F. Bernath J. Chem. Phys. 124, 156101 (2006) Spectroscopic Constants of BeH2 Gv-ZPE B Cal.Obs. Level Calc. 0 (0,0 ,0) 0.00 0.00 4.7023 0.0009 1.0528 0.0028 0 (0,0 ,1) 2178.79 -0.08 4.6333 0.0011 1.0360 0.0029 0 (0,0 ,2) 4323.80 0.02 4.5656 0.0014 1.0196 0.0026 0 (0,0 ,3) 6434.32 0.21 4.4989 0.0017 1.0040 0.0055 0 (1,0 ,0) a 0 (1,0 ,1) a+2120.28 1 (0,1 ,0) b 1 (0,1 ,1) b+2165.80 1 (0,1 ,2) a+4298.26 1 (1,1 ,0) c 1 (1,1 ,1) c+2106.97 a = 711.77 -1 cm Calc. Cal.Obs. 10**4 D Cal.Calc. Obs. 10**2 q Cal.Calc. Obs. 10**6 qD Cal.Calc. Obs. 4.6456 0.0012 1.0442 0.0018 0.12 4.5755 0.0014 1.0289 0.0029 4.7130 0.0010 1.0958 0.0050 -9.148 -0.008 8.33 0.21 -0.01 4.6440 0.0012 1.0795 0.0048 -9.107 -0.008 8.25 0.22 0.15 4.5762 0.0014 1.0634 0.0041 -9.069 -0.008 8.17 0.19 4.6558 0.0013 1.0853 0.0034 -9.010 0.012 8.07 -0.57 4.5857 0.0015 1.0698 0.0029 -8.961 0.013 7.96 -0.63 0.18 b = 1991.97 -1 cm c = 2702.78 -1 cm BeH2 for levels involving L-type resonance Gv-ZPE Level Calc. 0 (0,2 ,0) d 2 (0,2 ,0) d+10.61 B Cal.Obs. Calc. Cal.Obs. 10**4 D Cal.Calc. Obs. 10**2 q Cal.Calc. Obs. 10**6 qD Cal.Calc. Obs. 4.7247 0.0010 1.1442 0.0046 0.05 4.7225 0.0010 1.1378 0.0041 -9.203 -0.011 8.63 0.26 0 (0,2 ,1) d+2152.83 0.02 4.6557 0.0013 1.1284 0.0052 2 (0,2 ,1) d+2163.34 0.09 4.6534 0.0012 1.1221 0.0038 -9. 158 -0.010 8.55 0.22 0 (0,2 ,2) d+4272.76 0.21 4.5878 0.0015 1.1127 0.0023 2 (0,2 ,2) d+4283.18 0.30 4.5856 0.0015 1.1062 0.0050 8.50 0.19 d = 1417.27 -9.117 -0.011 -1 cm The RMS of discrepancies for 11 bands 0.15(±0.09) Relative inertial rotational constants B only 0.028% -1 cm Spectroscopic Constants of BeD2 Gv-ZPE B Cal.Obs. Cal.Obs. Level Calc. 0 (0,0 ,0) 0.00 0.00 2.36034 -0.0006 2.62478 0.0042 0 (0,0 ,1) 1689.27 -0.40 2.32970 -0.0006 2.57460 -0.0154 0 (0,0 ,2) 3356.06 -0.68 2.29950 -0.0004 2.52583 0.0093 1 (0,1 ,0) a 1 (0,1 ,1) a+1680.20 0 (0,2 ,0) d 2 (0,2 ,0) d+10.14 -0.36 Calc. 10**4 D Cal.Calc. Obs. -2.985 -0.033 1.31 -0.59 2.33655 -0.0005 2.67047 -0.0006 -2.995 -0.035 1.31 -0.71 -2.988 -0.009 1.36 0.16 -2.996 -0.010 1.36 0.20 2.37421 -0.0007 2.82248 0.0182 2.37370 -0.0005 2.81084 0.0539 0 (0,2 ,1) d+1671.17 -0.33 2.34356 -0.0006 2.77247 0.0142 2 (0,2 ,1) -0.29 2.34305 -0.0003 2.75870 0.0688 a = 548.36 -1 cm 10**6 qD Cal.Calc. Obs. 2.36720 -0.0006 2.72097 0.0276 -0.03 d+1681.26 10**2 q Cal.Calc. Obs. d = 1089.19 -1 cm The RMS of discrepancies for 5 bands 0.46(±0.19) Relative inertial rotational constants B only 0.023% -1 cm Predicted Spectroscopic Constants of BeHD ℓ (v1,v2 ,v3) Gv-ZPE B 10**5 D 10**9 H 0 (0,0 ,0) 0.00 3.2324 5.2446 0.95 0 (0,0 ,1) 1537.76 3.1948 5.2112 0.97 0 (0,0 ,2) 3045.66 3.1571 5.1685 1.04 0 (1,0 ,0) 2106.14 3.1943 5.1523 0.83 0 (2,0 ,0) 4153.26 3.1568 5.0297 0.88 1 (0,1 ,0) 635.55 3.2402 5.4226 1 (0,1 ,1) 2173.56 3.2031 1 (1,1 ,0) 2732.42 0 (0,2 ,0) 10**2 q 10**6 qD 10**10 qH 1.10 -4.860 3.124 -2.41 5.3597 1.19 -4.755 3.051 -2.60 3.2017 5.3329 1.00 -4.961 2.929 -3.28 1262.95 3.2483 5.6362 1.23 2 (0,2 ,0) 1275.16 3.2471 5.5912 1.25 -4.867 3.297 -2.82 0 (0,2 ,1) 2800.20 3.2114 5.5641 1.34 2 (0,2 ,1) 2813.07 3.2104 5.5069 1.35 -4.765 3.218 -2.62 0 (1,2 ,0) 3350.67 3.2098 5.5456 1.15 2 (1,2 ,0) 3362.64 3.2085 5.5021 1.14 -4.954 3.130 -3.32 Future work Difference Vtot-VH-H vs. rs=R1,(BeH)+R2,(BeH) & rd=R1,(BeH)-R2,(BeH) Trial Analytic Function V (rs , rd , ) U e {e 2 ( rs ;rd , )( rs rse ) 2e ( rs ;rd , )( rs rse ) } Conclusions • 3D PES constructed from 6864 ab initio points at icMRCI/AV5Z level and core correlation corrected at MR-ACPF/cc-pCV5Z. • Calculated spectroscopic constants of BeH2 and BeD2 are in excellent agreement with recent experiment results validating the quality of PES. • The predicted spectroscopic constants of BeHD should provide a reliable information for the unobserved bands. • The effect of different interpolation methods has been examined. • Future work Acknowledgments • P. F. Bernath, A. Shayesteh and M. Nooijen (University of Waterloo) • K. A. Peterson (Washington State University) • NSERCC Structures for (BeH2)n polymers G. J. Brendel et al. Inorg. Chem. 17, 3589(1978) X. F. Wang et al. Inorg. Chem. 44, 610(2005) Crystal structure of BeH2 showing a network of corner-sharing BeH4 tetrahedra G. S. Smith et al. Solid State Communications 67, 491(1988) Spectroscopic constants (in a V 0 (1,0 ,0) b V+C of BeH2 at different levels d V+C+F+R f V+C+F+R+B This work Exp. 1992.07 1992.18 1991.48 1991.97 716.65 716.24 716.36 716.24 716.48 0 (0,0 ,1) 2170.62 2180.47 2179.28 2179.39 2178.70 2178.79 2178.8659 B000 4.67506 4.70681 4.70560 4.70629 4.70407 4.70230 4.70140 4 10 D000 1.04192 1.05404 1.05473 1.05507 1.0543 1.0528 1.0500 1 g (0,1 ,0) 1984.77 1993.58 c V+C+F -1 cm ) 713.41 Reference to Jacek Koput and Kirk A. Peterson J. Chem. Phys. (2006) a • Determined using the CCSD(T) method with the cc-Pv6z and aug-cc-Pv6z basis sets for Be and H, respectively. b • Including additional corrections for the core-related correlation effects calculated at the CCSD(T)/cc-pCV5Z level. c • Including additional corrections for the high-order valence correlation effects calculated at the FCI/ and CCSD(T)/ cc-pCVTZ levels. d • Including additional corrections for the scalar relativistic effects calculated at the CCSD(T)/ cc-pV5Z-DK level. f • Including Born-Oppenheimer diagonal correction calculated CISD/ cc-pCVTZ( augcc-pVTZ on H) f • J=L=1 DVR and Lanczos algorithm • A direct product discrete variable representation (DVR) grid was used in vibrational energy calculation Each stretching coordinate was represented by 70 PODVR grid, with 200 equidistant sine-DVR grid on interval [1.6, 5.0] 80 Gauss-Legendre grid points on the interval [60-180] were used for the angular variable • Lanczos recursion used to diagonalize the Hamiltonian ˆ m1 [ m1m1 ( m )m ] 1 m ˆ m (m Hm m1m1 ), ˆ ˆ m ( m1m1 ( m )m m1m1 ( m )m ) 1/ 2 0 0 When eigenfunctions are needed, for selected eigenvalue , using inverse iteration method get the eigenvector and repeat Lanczos recursion get the wavefunction 10000 Lanczos iterations were found adequate to converge the levels within 9000 cm 0.001cm Remove Spurious Eigenvalues Methods • The Cullum-Willoughby (CW) method compare the eigenvalues of its sub-matrix obtained by deleting the first row and first column. If a Lanczos eigenvalue appears for both matrices, it is regarded “spurious” and deleted. • Different Recursion Steps Compare the Lanczos eigenvalues in different recursion steps: a true eigenvalue should not depend on recursion steps. With earliest appearance in the recursion is regarded as “good”. Since the “spurious” eigenvalues are shared by both the Lanczos matrix and its submatrix, they have small z1i and can be considered as copies generated from the round-off errors. J. Cullum and R. A. Willoughby, J. Comput. Phys. 44, 329 (1981) R. Chen and H. Guo J. Chem. Phys. 111, 9944 (1999) Triatomic Radau coordinaties O: Be atom m1=m2=H D: Central of mass of two H C: triatom central of mass 2 B: BD =CD·OD B. R. Johnson et al. J. Chem. Phys. 85,15(1986)