gradient	g1	g2	g3	g4
strength [%]	30	50	±80	8.1

¹⁵N decoupler attenuation and 90° pulse for GARP [200 μs, 19 dB]
initial value for t₁ evolution: 3 μs
initial value for t₂ evolution: 3 μs
increment for t₁ evolution: 1/[2·sw1]
increment for t₂ evolution: 1/[2·sw2]
ds: 32
ns: 16

5. Processing

Apply zero-filling to 128 real data points in F₂ and to 256 real data points in F₁ to obtain a matrix of 1024 × 128 × 256 real data points. This would result in a huge file of processed real data. Since only the amide proton region is of interest, use strip transformation in F₃ (400 points). Application of forward linear prediction (20 coefficients) for both F₂ and F₁ results in better resolved 3D spectra. Use Gaussian multiplication [gb = 0.1, lb = −3 Hz] in F₃ and a π/3-shifted squared sinusoidal window in the other dimensions for assignment purposes. The back lobes created by this window function may, however, cause problems for integration. For the data file described, extensive experimentation with different window functions may be worthwhile. Phase correction and base-line correction may be necessary for all dimensions. Further details are very dependent on the particular software you use to process such a 3D data file. Reference the ¹⁵N dimension (F₂) using the Ξ-scale procedure described in the introduction to this chapter using the DSS signal set to δ_H = 0.

6. Result

The figures show results from the 3D spectrum obtained on a DRX-600 spectrometer with a multinuclear inverse z-gradient triple resonance probe-head (fixed third channel on ¹³C). To demonstrate the huge amount of information, we first show in a the projection along the 1,3 planes of the cuboid. All these NOE cross-peaks must be disentangled, individually assigned, and integrated. Although the assignment of all peaks should have been done using the triple-resonance sequences discussed earlier in this chapter, the use of this information for the 3D NOESY spectrum measured in this experiment is by no means easy and straightforward, since the resolution in the indirect proton dimension is limited, and the ¹³C frequency which helped in achieving signal separation in the other 3D sequences is not present (cf. Exp. 15.19). In a first step, one therefore assigns only non overlapping and clearly identifiable signals, and in a later time-consuming procedure, iteratively guided by structure calculations, one tries to assign the ambiguous NOEs.

As an example we show the H,H strips of the amino-acid pair alanine A-46 and glycine G-47, taken from the corresponding planes at δ_N = 132.2 (b) and 102.3 (c). Figure b shows the diagonal peak for A-46 at δ_H = 8.97, and Figure c the diagonal peak for G-47 at δ_H = 8.12. In both figures the NH-NH NOE contact can be seen.