Advertisement

Journal of Molecular Modeling

, 25:257 | Cite as

Correction to: The Feynman dispersion correction for MNDO extended to F, Cl, Br and I

  • Maximilian Kriebel
  • Andreas Heßelmann
  • Matthias Hennemann
  • Timothy ClarkEmail author
Correction
  • 187 Downloads

Correction to: Journal of Molecular Modeling (2019) 25: 156

  https://doi.org/10.1007/s00894-019-4038-z

A small coding error in the development version of EMPIRE led to some inconsistencies in the above article. They are corrected in this erratum.

Introduction

Our recent description of MNDO-F for the halogens [1] was affected by a small coding error that was introduced into the code after the original MNDO-F article [2] and resulted in small deviations from the correct result. As this error was also present during the parameterization, we now present the correct parameters and results.

Results

Parameterization

Reparameterization using the correct code led to slightly better results for the training set, as shown in Fig. 1 and in the Supporting Information.
Fig. 1

Comparison of reference and MNDO-F interaction energies for the entire training set of C, H, N, O, F, Cl, Br and I compounds. The solid red line indicates the best-fit regression

The correct parameters are shown in Table 1:
Table 1

The optimized parameters for MNDO-F for H, C, N, O, F, Cl, Br and I

Element

Original MNDO

MNDO-F

RMSD

(kcal mol-1)

αA

αA

kA (bohr6)

rBJ (bohr)

H

2.544134

2.746420 (2.651305)

18.1663804 (22.749633)

3.5635804 (4.0216)

0.311

C

2.546380

2.640525 (2.58139)

7.752681 (7.971853)

5.385236 (4.0216)

N

2.861342

3.031761 (3.012587)

3.564848 (4.387597)

4.425983 (4.0216)

O

3.160604

3.125117 (3.201357)

8.823789 (7.416029)

3.644720 (4.0216)

 

F

3.419661

3.228403

3.094361

3.538267

0.294

Cl

2.542201

2.575831

5.691735

4.341654

0.209

Br

2.44570512

2.331572

6.717021

4.627255

0.354

I

2.20732001

1.754901

29.413381

4.664481

0.525

 

All halogen compounds

0.337

 

Entire Dataset

0.313

The values in parentheses are those given in reference [2] for a single rBJ value for H, C, N and O

The performance for the training set is marginally better than that reported in reference [1] with the exception of iodine compounds, which are slightly worse. The optimized parameters, however, have changed significantly, so that the correlation between the Feynman coefficient and the cube of the atomic polarizability reported in reference [1] is no longer as good. The detailed results are shown in the ESM.

Test complexes

Complexes between 1,3,5-trihalo- and perhalobenzenes and benzene

Table 2 and Fig. 2 show comparisons of MP2, DFT-SAPT and MNDO-F complexation energies for these complexes.
Table 2

Calculated complexation energies for 1,3,5-trihalogenated and perhalogenated benzenes complexed (π-π) to benzene

  

MP2-cp

DFT-SAPT

MNDO-F

C6H3Cl3:C6H6

(staggered)

-7.72

-4.22

-5.59

C6H3Cl3:C6H6

(eclipsed)

-7.55

-4.23

-5.58

C6Cl6:C6H6

(staggered)

-12.25

-6.21

-7.58

C6Cl6:C6H6

(eclipsed)

-11.8

-6.29

-7.58

C6H3Br3:C6H6

(staggered)

-8.4

-3.95

-6.66

C6Br6:C6H6

(staggered)

-13.19

-5.08

-9.79

C6H3I3:C6H6

(staggered)

-9.24

-4.52

-9.83

C6I6:C6H6

(staggered)

-14.58

-7.04

-16.08

Complexation energies are given in kcal mol-1 and abbreviations are those uses in reference [1], Table 2

Fig. 2

A comparison of the MP2/aug-cc-pVTZ, DFT-SAPT (CBS) and MNDO-F interaction energies for the π-π complexes of polyhalogenated benzenes with benzene

Halogen-bonded complexes

Figure 3 shows the correct results for the complexes of C6H5X with methylamine shown in Fig. 4 of reference [1].
Fig. 3

A comparison of the reference and MNDO-F interaction-energy curves for the halogen-bonded complexes of C6H5X (X = Cl, Br, I) with trimethylamine

Fig. 4

MNDO and MNDO-F calculated Heats of Formation (kcal mol-1) compared to experiment. The isolated point is C60-fullerene. The MNDO and MNDO-F regression lines and equations are shown

The performance is generally similar to that reported in reference [1] except that the energy profile for iodobenzene is considerably improved.

All other test-set results are shown in the ESI.

Heats of formation

Table 3 shows the MNDO and MNDO-F values for Eisol (eV) and the heats of formation of the atoms (kcal mol-1). The individual results are shown in the ESI and graphically in Fig. 4.
Table 3

MNDO and MNDO-F parameters for heats of formation

Element

Eisol (eV)

ΔHatom (kcal mol-1)

MNDO

MNDO-F

H

-11.906276

-12.58265

52.102

C

-120.500606

-122.22010

170.89

N

-202.566201

-205.21905

113.00

O

-317.868506

-319.63700

59.559

F

-476.683781

-476.86600

18.86

Cl

-353.1176787

-353.71210

28.99

Br

-346.68144152

-346.63050

26.74

I

-340.5984282

-339.91580

25.517

Eisol is the total energy of the isolated atom and ΔHatom its Heat of Formation. Eisol is a variable parameter in MNDO-F. The numbers of significant figures are those used in the EMPIRE program

The results are comparable to those reported in reference [1] (slightly worse than MNDO) but the excellent agreement found previously for C60-fullerene is no longer as good.

Dipole moments

Figure 5 show a comparison of MNDO and MNDO-F calculated dipole moments with experiment.
Fig. 5

Errors (calculated – experimental dipole moment, Debye) in MNDO and MNDO-F calculated dipole moments

Conclusions

The effect of the small program error present in reference [1] has been corrected. In general, MNDO-F performs as well or better than reported previously. The parameters reported in this erratum should be used for MNDO-F calculations.

The correct version of MNDO-F is available in the newest version of EMPIRE [3, 4, 5].

Notes

Acknowledgements

This work was supported by the Deutsche Forschungsgemeinschaft (DFG) Projektnummer 182849149 – SFB 953 “Synthetic Carbon Allotropes” and the Priority Program 1928 “Coordination Networks: Building Blocks for Functional Systems” (Cl85/24-1), and by the Bayerische Staatsministerium für Wissenschaft, Forschung und Kunst as part of the “Solar Technologies Go Hybrid” initiative.

Supplementary material

894_2019_4142_MOESM1_ESM.pdf (644 kb)
ESM 1 (PDF 643 kb)

References

  1. 1.
    Kriebel M, Heßelmann A, Hennemann M, Clark T (2019) The Feynman dispersion correction for MNDO extended to F, Cl, Br and I. J. Mol. Model. 25:156.  https://doi.org/10.1007/s00894-019-4038-z CrossRefPubMedGoogle Scholar
  2. 2.
    Kriebel M, Weber K, Clark T (2018) A Feynman dispersion correction: a proof of principle for MNDO. J. Mol. Model. 24:338.  https://doi.org/10.1007/s00894-018-3874-6 CrossRefPubMedGoogle Scholar
  3. 3.
    Hennemann M, Clark T (2014) EMPIRE: a highly parallel semiempirical molecular orbital program: 1: self-consistent field calculations. J. Mol. Model. 20:2331.  https://doi.org/10.1007/s00894-014-2331-4 CrossRefPubMedGoogle Scholar
  4. 4.
    Margraf JT, Hennemann M, Meyer B, Clark T (2015) EMPIRE: a highly parallel semiempirical molecular orbital program: 2: periodic boundary conditions. J. Mol. Model. 21:144.  https://doi.org/10.1007/s00894-015-2692-3 CrossRefPubMedPubMedCentralGoogle Scholar
  5. 5.
    Cepos InSilico GmbH (2018) EMPIRE software. Cepos InSilico GmbH,Erlangen. http://www.ceposinsilico.de/products/empire.htm. Accessed 16th July 2019

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2019

Authors and Affiliations

  1. 1.Computer-Chemistry Center, Department of Chemistry and PharmacyFriedrich-Alexander-University Erlangen-NürnbergErlangenGermany
  2. 2.Chair of Theoretical Chemistry, Department of Chemistry and PharmacyFriedrich-Alexander-University Erlangen-NürnbergErlangenGermany

Personalised recommendations