Tutorial on Energetic Analysis with NBO Deletions ($DEL Keylist)

Introduction to the $DEL Keylist and NBO Energetic Analysis

When interfaced to Gaussian or GAMESS, the NBO program makes it possible to determine the energetic effect of deleting certain NBOs, groups of NBOs, or specific NBO donor-acceptor interactions. This is done by including (following the main $NBO keylist) a special "deletions" $DEL keylist of the form

$DEL keyword(s) $END

[For Gaussian, the POP=NBO6DEL keyword (for Rev D) or POP=NBODEL keyword (for pre-Rev D) must also be included on the route card.]

The NBO Manual (Section B.5) describes nine deletion types (each with specific keyword syntax) that can be used to select specific classes of intra- or intermolecular Fock matrix elements for deletion. When combined with geometry optimization, the $DEL keylist also makes it possible to determine the structural consequences of specific NBO donor-acceptor interactions as the difference between the original fully-optimized geometry and the $DEL-optimized geometry.

In this Tutorial we employ $DEL keyword options to illustrate how simple chemical questions may be addressed in the case of a configurational energy difference.

Cis vs. Trans Configuration of Difluoroethene

1,2-difluoroethene exists in cis and trans isomers:



Of the two, the trans isomer is expected to be significantly more stable, on the basis of steric and electrostatic factors, because the two bulky F atoms and the repelling negative charges of the two CF bond dipoles are maximally separated in this configuration. However, the cis isomer is actually found to be very similar in energy. For example, for a simple RHF/6-31G* model with idealized trigonal geometry (RCF = 1.33°, RCC = 1.34°, RCH = 1.08°) the calculated energies are

Ecis = -275.717264 a.u.,
Etrans = -275.717614 a.u.

corresponding to only a slight energy difference favoring the trans isomer.

Ecis - Etrans = 0.22 kcal/mol

The model RHF calculation exhibits the near-equivalence of cis and trans configurational energies but does not help us to understand why the cis configuration is so competitive. In this Tutorial we show how to address such questions by using $DEL keylists to investigate the effect of specific NBO donor-acceptor interactions on the isomeric energy difference.

Can the Surprising Stability of the Cis Isomer be Attributed to Electronic Delocalization?

The role of electronic delocalization can be quantitatively assessed by deleting all non-Lewis NBOs from the basis set with the LEWIS keyword

$DEL LEWIS $END

The resulting "natural Lewis structure" wavefunction is perfectly localized, with all Lewis-type NBOs doubly occupied. By the variational principle, the Lewis-type wavefunction has an energy E(L) that is higher than the original energy E(full). The net energy difference E(NL)

E(NL) = E(full) - E(L)

gives the stabilizing effect of the delocalizing (non-Lewis) contributions.

For the cis isomer, the sample Gaussian input file

# rhf/6-31g* pop=nbo6del nosymm

cis-1,2-difluoroethene...rhf/6-31g*//idealized geometry

0 1
C
C  1  1.34
F  1  1.33  2  120.
F  2  1.33  1  120.  3    0. 0
H  1  1.08  2  120.  3  180. 0
H  2  1.08  1  120.  4  180. 0

$nbo  $end
$del  lewis  $end

leads to the $DEL output shown below. The first output segment identifies the type of deletion ("LEWIS: Delete all non-Lewis NBOs") and lists the 48 antibond and Rydberg NBOs (17-64) that are deleted:

 LEWIS: Delete all non-Lewis NBOs
 Deletion of the following orbitals from the NBO Fock matrix:
   17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36
   37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56
   57  58  59  60  61  62  63  64

The next segment lists the NBOs with their original occupancies ("No deletions") and current occupancies ("This deletion"), together with the net change in each:

 Orbital occupancies:

       Orbital                         No deletions  This deletion    Change
 ------------------------------------------------------------------------------
    1. CR ( 1) C  1                      1.99872        2.00000       0.00128
    2. CR ( 1) C  2                      1.99872        2.00000       0.00128
    3. CR ( 1) F  3                      1.99994        2.00000       0.00006
    4. CR ( 1) F  4                      1.99994        2.00000       0.00006
    5. LP ( 1) F  3                      1.99225        2.00000       0.00775
    6. LP ( 2) F  3                      1.97329        2.00000       0.02671
    7. LP ( 3) F  3                      1.94905        2.00000       0.05095
    8. LP ( 1) F  4                      1.99225        2.00000       0.00775
    9. LP ( 2) F  4                      1.97329        2.00000       0.02671
   10. LP ( 3) F  4                      1.94905        2.00000       0.05095
   11. BD ( 1) C  1- C  2                1.99758        2.00000       0.00242
   12. BD ( 2) C  1- C  2                1.99676        2.00000       0.00324
   13. BD ( 1) C  1- F  3                1.99680        2.00000       0.00320
   14. BD ( 1) C  1- H  5                1.98730        2.00000       0.01270
   15. BD ( 1) C  2- F  4                1.99680        2.00000       0.00320
   16. BD ( 1) C  2- H  6                1.98730        2.00000       0.01270
   17. BD*( 1) C  1- C  2                0.02163        0.00000      -0.02163
   18. BD*( 2) C  1- C  2                0.09741        0.00000      -0.09741
   19. BD*( 1) C  1- F  3                0.00987        0.00000      -0.00987
   20. BD*( 1) C  1- H  5                0.01756        0.00000      -0.01756
   21. BD*( 1) C  2- F  4                0.00987        0.00000      -0.00987
   22. BD*( 1) C  2- H  6                0.01756        0.00000      -0.01756
   23. RY ( 1) C  1                      0.00612        0.00000      -0.00612
   24. RY ( 2) C  1                      0.00453        0.00000      -0.00453
   25. RY ( 3) C  1                      0.00198        0.00000      -0.00198
   26. RY ( 4) C  1                      0.00131        0.00000      -0.00131
   27. RY ( 5) C  1                      0.00026        0.00000      -0.00026
   28. RY ( 6) C  1                      0.00017        0.00000      -0.00017
   29. RY ( 7) C  1                      0.00008        0.00000      -0.00008
   30. RY ( 8) C  1                      0.00001        0.00000      -0.00001
   31. RY ( 9) C  1                      0.00000        0.00000       0.00000
   32. RY (10) C  1                      0.00000        0.00000       0.00000
   33. RY ( 1) C  2                      0.00612        0.00000      -0.00612
   34. RY ( 2) C  2                      0.00453        0.00000      -0.00453
   35. RY ( 3) C  2                      0.00198        0.00000      -0.00198
   36. RY ( 4) C  2                      0.00131        0.00000      -0.00131
   37. RY ( 5) C  2                      0.00026        0.00000      -0.00026
   38. RY ( 6) C  2                      0.00017        0.00000      -0.00017
   39. RY ( 7) C  2                      0.00008        0.00000      -0.00008
   40. RY ( 8) C  2                      0.00001        0.00000      -0.00001
   41. RY ( 9) C  2                      0.00000        0.00000       0.00000
   42. RY (10) C  2                      0.00000        0.00000       0.00000
   43. RY ( 1) F  3                      0.00171        0.00000      -0.00171
   44. RY ( 2) F  3                      0.00159        0.00000      -0.00159
   45. RY ( 3) F  3                      0.00007        0.00000      -0.00007
   46. RY ( 4) F  3                      0.00004        0.00000      -0.00004
   47. RY ( 5) F  3                      0.00002        0.00000      -0.00002
   48. RY ( 6) F  3                      0.00001        0.00000      -0.00001
   49. RY ( 7) F  3                      0.00000        0.00000       0.00000
   50. RY ( 8) F  3                      0.00000        0.00000       0.00000
   51. RY ( 9) F  3                      0.00000        0.00000       0.00000
   52. RY (10) F  3                      0.00000        0.00000       0.00000
   53. RY ( 1) F  4                      0.00171        0.00000      -0.00171
   54. RY ( 2) F  4                      0.00159        0.00000      -0.00159
   55. RY ( 3) F  4                      0.00007        0.00000      -0.00007
   56. RY ( 4) F  4                      0.00004        0.00000      -0.00004
   57. RY ( 5) F  4                      0.00002        0.00000      -0.00002
   58. RY ( 6) F  4                      0.00001        0.00000      -0.00001
   59. RY ( 7) F  4                      0.00000        0.00000       0.00000
   60. RY ( 8) F  4                      0.00000        0.00000       0.00000
   61. RY ( 9) F  4                      0.00000        0.00000       0.00000
   62. RY (10) F  4                      0.00000        0.00000       0.00000
   63. RY ( 1) H  5                      0.00061        0.00000      -0.00061
   64. RY ( 1) H  6                      0.00061        0.00000      -0.00061

Note that this deletion has resulted in all Lewis-type NBOs (1-16) becoming doubly occupied and all remaining non-Lewis-type NBOs (17-64) becoming empty, the idealized Lewis structure limit.

The next output segment displays the calculated energy of the deletion and the energy change (in atomic units and kcal/mol):

 NEXT STEP:  Perform one SCF cycle to evaluate the energy of the new density
             matrix constructed from the deleted NBO Fock matrix.

 ------------------------------------------------------------------------------
 .
 .
 .
 Requested convergence on RMS density matrix=1.00D-08 within   1 cycles.
 Requested convergence on MAX density matrix=1.00D-06.
 Requested convergence on             energy=1.00D-06.
 No special actions if energy rises.
 Fock symm off for IB=2 I1=    1 I=   25 J=   14 Cut=1.00D-07 Err=3.63D-02
 Fock matrix is not symmetric: symmetry in diagonalization turned off.
 >>>>>>>>>> Convergence criterion not met.
 SCF Done:  E(RHF) =  -275.461162154     A.U. after    2 cycles
            NFock=  1  Conv=0.92D-02     -V/T= 2.0040
 .
 .
 .
 ------------------------------------------------------------------------------
   Energy of deletion :       -275.461162154
     Total SCF energy :       -275.717263936
                          -------------------
        Energy change :          0.256102 a.u.,         160.706 kcal/mol
 ------------------------------------------------------------------------------

[Note that the Gaussian program always prints a "Convergence criterion not met" warning message, because the deleted density does not exactly match the starting density in the one-pass SCF evaluator method used by $DEL. This message can be safely ignored. For the LEWIS case, the energy of deletion is a strict variational expectation value of the natural Lewis structure wavefunction, but for other deletions a slight variational inconsistency is incurred due to the slight difference between the deleted density distribution and the final density from which the converged Fock operator was constructed.]

[Also note that the "nosymm" keyword is added to the route card of the Gaussian input file for deletions calculation to override the use of symmetry in the SCF energy evaluation. This keyword can be safely removed if the deletions to be performed ensure that the deleted density retains the symmetry of the fully delocalized SCF density. In fact, the Lewis deletions described here do preserve symmetry, so "nosymm" could be discarded. However, several of the deletions described below do not preserve symmetry. If uncertain, include "nosymm" on the Gaussian route card. Additional comments on symmetry can be found in Section B.5 of the NBO Manual.]

From the above output, we recognize that

Ecis(full) = -275.717264 a.u.
Ecis(L) = -275.461162 a.u.
Ecis(NL) = -0.256102 a.u.

Similarly, from the $DEL output for the trans isomer, we obtain

Etrans(full) = -275.717614 a.u.
Etrans(L) = -275.466245 a.u.
Etrans(NL) = -0.251369 a.u.

From these values we find

Ecis(L) - Etrans(L) = +0.005083 a.u. = +3.19 kcal/mol
Ecis(NL) - Etrans(NL) = -0.004733 a.u. = -2.97 kcal/mol

Thus, the localized E(L) contribution favors the trans configuration by >3 kcal/mol, in accordance with the expected steric and electrostatic differences. However, the small delocalization contribution is seen to strongly favor the cis isomer by a similar amount, leading to nearly equal total energies for the two isomers. The surprising stability of the cis isomer can therefore be attributed to the electronic delocalization energy E(NL).

What Specific NBO Donor-Acceptor Interactions are Responsible for this Preference?

To further dissect E(NL) into specific donor-acceptor interactions, we can consider more selective $DEL keylists of the form

$DEL
delete n elements
  d1 a1
  d2 a2
   . . .
  dn an
$END

where each di is the number of a Lewis-type donor NBO and ai is the number of a non-Lewis-type acceptor NBO. For example, the possible vicinal hyperconjugative delocalizations between the CF bond (NBO 13) and CH bond (NBO 14) on C(1) with the CF antibond (NBO 21) and CH antibond (NBO 22) on C(2) could be specified with

$DEL
delete 4 elements
  13  21
  13  22
  14  21
  14  22
$END

By such exploratory deletions, one can verify that donor-acceptor interactions involving the Rydberg-type (RY) NBOs 23-64 have negligible effect on the cis/trans energy difference. Similarly, interactions involving the acceptor antibond (BD*) NBOs 17-22 are only significant when paired with vicinal donor bond (BD) NBOs on the opposite carbon atom.

The most important contributions to the cis-trans energy difference are found to be the two equivalent vicinal interactions involving CH bonds as donors (NBOs 14,16) and CF* antibonds as acceptors (NBOs 19,21). The 14-21 interaction, e.g., can be deleted with the $DEL keylist

$DEL
delete 1 element
  14  21
$END

The deletion energy for each such CH-CF* interaction is found to be 5.97 kcal/mol in the cis isomer but only 1.19 kcal/mol in the trans isomer, thus contributing strongly to stabilizing the cis isomer. (The calculated deletion energies agree sensibly with the corresponding 2nd-order perturbative estimates of 6.86 and 1.40 kcal/mol.)

Why do vicinal CH-CF* interactions favor the cis isomer? Each such CH-CF* interaction is antiperiplanar in the cis isomer, but synperiplanar in the trans isomer. As simple NBO overlap diagrams make clear, this delocalizing interaction is considerably stronger in the anti arrangement than in the syn.



Thus, the cis isomer benefits from two strong antiperiplanar CH-CF* delocalizations, whereas the trans isomer has none. [There is evidently a similar syn/anti difference between CH-CH* interactions that favors the trans isomer, but the CH-CF* interactions are the more important due to the greater acceptor strength of the CF* antibond.] We can therefore conclude that vicinal CH-CF* donor-acceptor interactions are most responsible for stabilizing the cis isomer.

What Influences Do Hyperconjugative Delocalizations Exert on Other Geometrical Variables?

From the NBO overlap diagrams shown above, we can visualize how slight distortions of geometry might further strengthen the donor-acceptor interactions. In the cis isomer, for example, opening up the C-C-F angle or closing down the C-C-H angle will both increase the favorable overlap between the backside lobe of the CF* antibond and the shoulder of the CH bond to allow stronger delocalization and stabilization, as will shortening the central C-C bond. Such distortions would naturally be opposed by the skeletal bonding framework, as the bonding hybrids strive to maintain optimal orientations and separations for strong covalent bonding. Given the weakness of hyperconjugative donor-acceptor delocalizations compared to covalent bonding interactions, we expect that the hyperconjugative effects on molecular geometry are small but detectable (perhaps a few degrees or a few hundredths of an Angstrom). How can we quantitate these hyperconjugative distortions?

A $DEL keylist combined with geometry optimization allows one to determine what the optimal geometry would be in the absence of a particular hyperconjugative interaction. As a simple example, let us consider how the geometry of difluoroethene would be altered if the two CF* antibonds were not present, so that no hyperconjugative delocalizations involving these orbitals were possible. We can address this question by re-optimizing the geometry with the two CF* antibonds (NBOs 19,21) deleted. The G09 input file to perform such a $DEL-optimization for the cis isomer is shown below. (The corresponding input file for GAMESS is given at the end of this tutorial.)

# rhf/6-31g* pop=nbo6del nosymm opt

Geometry optimization of cis-1,2-difluoroethene (no CF*s)

0 1
C
C  1  CC
F  1  CF  2  CCF
F  2  CF  1  CCF  3    0. 0
H  1  CH  2  CCH  3  180. 0
H  2  CH  1  CCH  4  180. 0

CC     1.34
CF     1.33
CH     1.08
CCF  120.
CCH  120.

$nbo  $end
$choose
  lone 3 3 4 3 end
  bond d 1 2 s 1 3 s 1 5 s 2 4 s 2 6 end
$end
$del  delete 2 orbitals 19 21  $end

In this job, the Gaussian geometry optimizer searches for an equilibrium structure on the E($DEL) surface, which differs from the E(full) surface by deletion of contributions from NBOs 19,21. Removing the POP=NBO6DEL keyword and repeating the optimization yields the usual full-basis optimized structure for comparison. Some bond angles and distances for the E(full) and E($DEL) optimized geometries are shown in the table below:

Optimized Geometry of Cis Isomer

Geometry Parameter full $DEL diff
bond length (Å) C-C 1.307 1.342 -0.035
C-F 1.324 1.308 +0.016
C-H 1.070 1.070 0.000
bond angle (°) C-C-F 122.6 117.9 +4.7
C-C-H 123.0 127.8 -4.8

From these comparisons one can see that delocalizations into the CF* antibonds have three primary effects on molecular geometry: (i) significant shortening (by 0.035 A) of the central C-C bond length; (ii) significant opening (by 4.7 degrees) of the C-C-F bond angle; (iii) significant closing (by 4.8 degrees) of the C-C-H bond angle. As noted above, all three of these changes are expected to enhance CH-CF* delocalization by strengthening donor-acceptor overlap.

Similar optimizations of the trans isomer reveal that CF* delocalizations shorten the C-C bond length but have comparatively limited influence on the C-C-F and C-C-H bond angles:

Optimized Geometry of Trans Isomer

Geometry Parameter full $DEL diff
bond length (Å) C-C 1.307 1.340 -0.033
C-F 1.329 1.308 +0.021
C-H 1.070 1.072 -0.002
bond angle (°) C-C-F 120.2 121.0 -0.8
C-C-H 125.3 124.5 +0.8

[Note that we recommend using the directed NBO search ($CHOOSE input) when performing geometry optimizations with deletions. These optimizations are based on numerical gradients, and thereby sample a large number of geometries as E($DEL) is minimized. $CHOOSE input ensures that NBO uses the same Lewis structure at each geometry. Because the orbital numbers used to specify the deletion are tied directly to a particular Lewis structure, changes in the Lewis structure from geometry to geometry are bound to confuse the optimizer. $CHOOSE input is described in Section B.4 of the NBO Manual.]

The following is the corresponding input file for GAMESS for 1,2-cis-difluoroethene.

 $contrl  scftyp=rhf  runtyp=trudge  coord=hint  $end
 $basis  gbasis=n31  ngauss=6  ndfunc=1  $end
 $trudge  optmiz=geometry  npar=5  iex(1)=11,21,22,31,32  $end
 $trurst  tolf=0.0001  tolr=0.001  fnoise=0.00001  $end
 $nbo  $end
 $choose
   lone 3 3 4 3 end
   bond d 1 2 s 1 3 s 1 5 s 2 4 s 2 6 end
 $end
 $del  delete 2 orbitals 19 21  $end
 $data
Geometry optimization of cis-1,2-difluoroethene (no CF*s)
Cnv 2

C    6.    LC    0.67     0.00    0.00 + O I
F    9.    TCT   1.33   120.00    0.00 + 1 O K
H    1.    TCT   1.08   120.00  180.00 + 1 O K
 $end

Note that the calculation uses GAMESS' TRUDGE numerical optimizer and that the geometry must be specified in Hildebrandt (coord=hint) internal coordinates. See the GAMESS manual for information on runtyp=trudge, coord=hint, and the $trudge and $trurst groups. GAMESS optimizes the cis isomer to the same $DEL geometry described above.


NBO Tutorials
NBO Home