Computation of Through-Space NMR Chemical Shift Effects

Ned H. Martin Department of Chemistry University of North Carolina at Wilmington

Introduction to NMR
In a strong magnetic field Bo, hydrogen nuclei have two possible spin states: aligned with or against the magnetic field. These states differ only slightly in energy. The energy difference between the two spin states corresponds (by Einsteins equation: E = hn) to energy in the radiofrequency region of the E = hn electromagnetic spectrum. When hydrogen nuclei are irradiated with the appropriate radiofrequency in a strong magnetic field, they absorb energy and spin-flip. This is NMR.

Bo

Introduction to NMR
Hydrogen nuclei are shielded from the full applied magnetic field Bo by the electrons surrounding them. Nearby electronegative atoms H H such as oxygen or chlorine O attract electrons, thus CH3 CH2 C CH2 O CH2Cl reducing electron density and causing deshielding H H of nearby hydrogens. Each type of hydrogen has a unique position 10 8 6 4 2 0 of absorption (called the chemical shift) in the NMR spectrum.

Estimation of Proton Chemical Shifts

Proton chemical shifts are usually estimated based on additive through-bond substituent effects. However, in some instances, through-space (shielding or deshielding) effects may be more important.
Observed deviations from estimated chemical shifts caused by through-space effects (- is shielding, + is deshielding):
Ha
Ha

Hb

Ha

Hb

+ 2.2 ppm

+ 0.2 ppm

- 0.3 ppm

- 3.1 ppm

Through-Space (de)Shielding
Structures that exhibit NMR shifts that deviate from those predicted by substituent effects generally have one or more protons near a p bond. Theoretical predictions of through-space magnetic effects resulted in the familiar shielding cones, such as the one for the C=C shown below.
+
shielding

q
deshielding

deshielding (q less than 54.7o)

shielding

Shielding Cones
Shielding cones are based on the McConnell equation

s = 1/3 c (1 - 3 cos2 q)/4pR3

which predicts the magnetic shielding increment at a point in space due solely to the magnetic anisotropy of the functional group (C=C in this case). Based on the McConnell equation, protons close to and over a C=C (within a 54.7 cone) should be Ha shielded (shifted upfield); in fact they are deshielded and McConnell s: +0.12 are shifted downfield.
Observed s: -2.12

Our Groups Reseach

Our groups research over the past eight years has focused on the use of ab initio quantum chemical calculations to: study through-space shielding effects of various functional groups Try to understand their origin and develop corrections to estimated shifts based on through-bond substituent effects.

Our Approach
Our approach has been to use ab initio computational methods to calculate isotropic shielding values of protons in simple model systems incorporating functional groups that exert through-space effects. For instance, to examine the effect of a C=C bond, our model system uses methane as a probe in various positions over ethene, the simplest molecule containing a C=C bond. Subtraction of the isotropic shielding value of protons in an isolated methane gives the shielding effect of the C=C functional group.

Methodology
The subroutine GIAO (gauge-including atomic orbital) in Gaussian was employed to calculate isotropic shielding values. H H HF/6-31G(d,p) calculations H C were performed on a simple H model system composed of previously-optimized H methane and ethene C H structures juxtaposed H C variously. H Symmetry reduced the number of calculations required.

Methodology
Single-point calculations were done on a series of supramolecules each having methane at a different position over the plane of ethene. This process was repeated at several distances (2.0, 2.5, 3.0, 3.5, 4.0, 4.5 and 5.0 ) between the proximate proton of methane and the plane of ethene. The isotropic shielding values of the proximate proton of methane were extracted from the Gaussian output. The isotropic shielding value of methane (by itself) calculated in the same way was subtracted from each of the above values. We define this difference as the shielding increment (s).

Methodology
The shielding increment (d) for each H position The was plotted against Cartesian coordinates to obtain a shielding surface at each distance above ethene.

2.0
1 0 -1 -2 -3 -4

3.0
0.2
s

0.1 0

1.5 Y

0.5

0 0
H

0.5

1.5 X

-0.1

1.5 Y

0.5

0 0

0.5

1.5 X

C H

Positive values are shielding; negative values are deshielding.

Methodology
A function was matched to each surface using TableCurve3D. The same form of mathematical function s = a + bX + cY + dX2 + eY2 + fXY . was found to give a good fit to the shielding surface at each of several distances of methane over ethene:
2.0 : s = 2.73 + 1.88X + 2.20Y 0.29X2 0.22Y2 0.77XY
2.5 : s = 0.79 + 0.82X + 0.70Y 0.22X2 0.14Y2 0.21XY 3.0 : s = 0.12 + 0.35X + 0.23Y 0.14X2 0.065Y2 0.038XY (etc., up to 5.0 )

Methodology
The values of the constant a and coefficients b, c, d, e & f in the equations of the form: s = a + bX + cY + dX2 + eY2 + fXY . varied smoothly as a function of distance. Each of these variables could be related to the distance above ethene using quadratic equations.
2 1.5 1 0.5 0 2 2.5 Z 3 3.5
b

2.5

y = 0.9162x 2 - 6.1398x + 10.478 R2 = 0.9979

c

2 1.5 1 0.5 0 2

y = 1.366x 2 - 8.8719x + 14.445 R2 = 0.991

2.5 Z

3.5

Methodology
Substitution of these quadratic equations into the general shielding surface equation resulted in one equation (18 terms; too big to show!) for predicting the through-space shielding increment as a function of Cartesian coordinates relative to the center of the C=C. This increment is useful as a correction for estimated chemical shifts.
Ha Hb

Ha

Hb
H

Fcn. Calc. s Obs. Dev. s

- 0.28 - 0.24

- 0.30 - 0.27

- 1.99 - 2.12

Discrepancy rel. to Mc Connell Eqn.

Our results show deshielding over the center of a C=C; the McConnell equation predicts shielding. McConnells equation considers only the magnetic anisotropy of the C=C (or other functional group); it disregards all other factors that affect the chemical shift!
H H C H H

H C H

H H

2 0 -2 0 s -4 -6 -8 -10

Distance above Ethene,

Martin et al., J. Am. Chem. Soc. 1998, 120(44), 11510-11511.

McConnell Eqn. vs. Our Function

Calcd NMR shielding increments along the C=C bond axis of ethene; (Blue = shielding; Red = deshielding)

3 Z 2.5 2

3 Z 2.5 2

-2 -1

-2 -1

Y McConnell Equation

Y Our Shielding Function

Martin et al., Int. J. Mol. Sci., 2000, 1, 84-91.

Results
We have reported results of such computational studies of the NMR shielding (or deshielding) surfaces over 1,2 and alkenes3 aromatic rings1,2 benzene rings and alkenes 3..

H H C H C H

1. Martin et al., J. Mol. Struct. (THEOCHEM) 1998, 454, 161-166. 2. Martin et al., J. Mol. Graphics Mod. 2000, 18(3), 242-246. 3. Martin et al., Struct. Chem. 1998, 9(6), 403-410, Struct. Chem. 1999, 10(5), 375-380, J. Molec. Graph. Mod. 2000, 18(1), 1-6.

Positive values are shielding; negative values are deshielding.

Results
The function-predicted through-space shielding increment compares favorably to the observed deviation from the estimated chemical shifts (based on additive substituent effects).
Observed deviations from estimated chemical shifts (Here, - is shielding, + is deshielding) and Function-computed chemical shift increments:
Ha
Ha

Hb
Ha Hb

+ 2.1 ppm + 2.0 ppm

+ 0.2 ppm + 0.3 ppm

- 0.3 ppm - 0.3 ppm

- 3.1 ppm - 3.0 ppm

Results
We have also reported on the NMR shielding surfaces of the ethynyl, cyano, and nitro groups using CH4 as a probe, with good prediction of chemical shift effects.
N H H C C H O C O N H

Chemical shifts Predicted: Adjusted:

8.7d 9.8d

Observed:

9.9d

10.9d

8.7d

8.7d 8.2d

10.3d

8.1d

Martin et al., J. Mol. Graphics Mod. 2002, 21, 51-56.

Results
Our most recently published NMR shielding surface study was of the carbonyl group, C=O. The traditional (McConnell) shielding cone picture of the carbonyl group in textbooks shows a cone of deshielding along the C=O bond axis, with shielding of deshielding along the C=O bond axis, with shielding above the C=O group. Our results differ differ:substantially:
shielding

C O
shielding

Martin et al., J. Mol. Graphics Mod. 2003, 22, 127-131.

Origin of NMR (de)Shielding Effects

In collaboration with P.v.R. Schleyer (U. Ga.), IGLO-HF was used to analyze the localized orbital origins of the through-space shielding effects due to the ethenyl, ethynyl, cyano, nitro and carbonyl groups. The results indicated that in each of these systems, the proximate C-H bond of the methane probe accounts for over 40% of the shielding increment. Thus, McConnells approach, based solely on magnetic anisotropy of the functional group can not be expected to predict chemical shift effects accurately.

Martin et al., Org. Lett. 2001, 3(24), 3823-3826.

Origin of (de)Shielding
The strong deshielding of a proton in the face of a C=C may be the result of mutual perturbation of the interacting orbitals of the probe and the test molecules. An indication of this is seen in the representation (right) of the HOMO of ethene (wiremesh) superimposed with the HOMO of a methane-ethene pair (solid), separated by 2.0
Martin et al., in Modeling NMR Chemical Shifts: Gaining Insights into Structure and Environment," ed. Facelli, J.C and deDios, A.C., ACS, Washington, D.C., 1999, 207-219.

Polarization of C-H Bond

Such a perturbation should be accompanied by a change in the calculated atomic charge. NPA charges were calculated for the proximal H of the probe methane over each functional group and also for the Hs on isolated methane. The difference between these was plotted vs. distance of the proximal H above ethene. Similar results were observed over ethyne; a similar pattern but with less charge difference was seen over HCN and over benzene.

Charge difference, Proximal H

0.06 0.05

Charge (e-)

0.04 0.03 0.02 0.01 0 1.5 2 2.5 3 3.5 4 Distance above Ethene,

Effect of Choice of Probe?

Several referees and researchers in this field have suggested using other probes, such as a ghost atom (Bq in Gaussian ), a H atom, or a He atom. It was also suggested that constrained geometryoptimized probe-test supramolecules (as opposed to the single point calculations we had performed) would give more accurate results. Our most recent work has involved investigating alternative computational probes of through-space NMR shielding effects to assess their validity and computational efficiency.

Methodology
The following probes were used to calculate the through-space shielding effect of several test molecules:

Bq (ghost atom) H atom (single point) H atom (geometry optimized) He atom (single point) He atom (geometry optimized) H2 molecule (single point) H2 molecule (geometry optimized) CH4 (single point) (the probe used in our previous work) CH4 (geometry optimized)

Methodology
The test molecules, simple structures containing common organic functional groups, included:
CH3 CH3 CH2 CH2 HC CH HC N

We are also examining the effect of the choice of probe over some small-ring hydrocarbons. Of these, only cyclopropane will be discussed today.
CH2 H2C CH2 CH2 HC CH H2C H2C CH2 CH2 H2C CH2 HC CH HC CH

HC CH

Single Point Calculations

Each HF/6-31G(d,p) geometry-optimized probe (Z) was placed over the HF/6-31G(d,p) geometryoptimized test structures in separate Cartesian coordinate input files. The probes position was moved 0.5 incrementally in the Z direction. Single point calculations were performed using GIAO in Gaussian 98.

Z
H H C C H H

Geometry-optimized calculations
Each HF/6-31G(d,p) geometryoptimized probe (Z) was placed over the HF/ 6-31G(d,p) geometry-optimized test structures in separate Z-matrix input files. A dummy atom X was placed at the reference point (here, the center of C=C bond). The distance between the probe and the dummy atom was fixed, but all other structural parameters were allowed to optimize.

Z
H H C XC H H

Z-Matrix Description of He over Ethene

0 1 X C1 X halfcc He X hX C1 C2 X halfcc He H1 C1 1.076 X H2 C1 1.076 X H3 C2 1.076 X H4 C2 1.076 X Variables: halfcc=0.65746 Constants: hX=2.0 a=90.0 b=180.0

a a 121.7 121.7 121.7 121.7

C1 b He1 90.0 He1 -90.0 He1 90.0 He1 -90.0

He

H H C XC H H

Shielding over the Center of the CarbonCarbon Single Bond in Ethane

2

Calculated shielding increment, ppm

-1

-2

Bq H H opt He He opt H2 H2 opt CH4 CH4opt

2 2.5 3 3.5

-3 Distance above C-C bond,

Shielding over the Center of the CarbonCarbon Double Bond in Ethene

2
Calculated shielding increment, ppm

1 0 -1 -2 -3 -4 -5 2 2.5 3 3.5 Distance from center of C=C,

Bq H H opt He He opt H2 H2 opt CH4 CH4 opt

Shielding over the Center of the CarbonCarbon Triple Bond in Ethyne

1 0
Calculated Shielding increment, ppm

-1

-2

-3

-4

Bq H H opt He He opt H2 H2opt CH4 CH4 opt

-5 2 2.5 3 Distance above center of C-C triple bond, 3.5

Shielding over the Center of the CarbonNitrogen Triple Bond in HCN

1
Calculated shielding increment, ppm

-1
Bq H H opt He He opt H2 H2 opt CH4 CH4 opt

-2

-3

-4 2 2.5 3 3.5 Distance above the C-N bond,

Shielding over the Center of the Benzene Ring

6 5

Calculated shielding increment, ppm

Bq H H opt He He opt H2 H2 opt CH4 CH4 opt

0 2 2.5 3 3.5 Distance above center of benzene,

CH4 Shielding Surface over Cyclopropane

NMR shielding increments over cyclopropane were computed using CH4 as a probe and using H2 as a probe in much the same way that the shielding increments over ethene and other models for functional groups were obtained. The resulting shielding increments were plotted vs. Cartesian coordinates to provide shielding surfaces.

H C H

H H

H H H H

CH4 Shielding Surface over Cyclopropane, 2.5

Top view

s
ds

0.6 0.2 -0.2 -0.6 -1 X 1 Y

2 -2

-1

Note vdW deshielding!

Positive (blue) is shielding; Negative (red) is deshielding.

Comparison of Probes over Cyclopropane, 2.5

CH4 probe H2 probe

CH4 vs. H2 Probe

The shielding surfaces obtained using the two different molecular probes are very similar. They differ slightly in the magnitude of shielding. The ratio of the isotropic shielding values for the two probes (H2 / CH4) was 0.84, regardless of the test molecule and independent of the position over the test molecule in the systems studied (ethane, ethene, ethyne, benzene, HCN). Both probes provide good agreement with experimental chemical shift effects in example structures. The H2 probe is considerably easier to employ and is more economical computationally.

Summary of Shielding Probes

Bq (ghost atom) gives the poorest agreement with observed chemical shift effects. It completely ignores the mutual perturbation of orbitals. Monatomic probes (H and He) are not much better. There is no appreciable difference between single-point calculations and geometry-optimized calculations. H2 and CH4 give generally similar results, with H2 providing isotropic shielding values 0.84 of those obtained with CH4. CH4 has been found previously to give accurate predictions of chemical shift effects of through-space shielding; H2 (with or without a minor correction) can do so also. H2 is simpler and cheaper computationally.

Ongoing Research
We have begun to study the NMR shielding surfaces of molecules and complexes of biochemical interest, modeling through-space NMR shift effects that are operative in peptides:
O H C O H O C OH H O C NH2 NH3 NH4

H H N H C O O 2N NO 2 CH3 CH3 CH3 NO 2

NH4

Mg

Acknowledgments
Student collaborators:
Noah W. Allen, III; Luong Vo; Jill C. Moore; Everett K. Minga; Sal T. Ingrassia; Justin D. Brown; H. Lee Woodcock; David M. Kmiec, Jr.; Kimberly H. Nance; Dustin C. Wade; David M. Loveless; Kristin L. Main. The donors of the American Chemical Society Petroleum Research Fund for support of this research (1996-2003)

The (former) North Carolina Supercomputing Center The UNCW Information Technology Services Division The UNCW College of Arts and Sciences The UNCW Department of Chemistry