Pure Shift NMR


Unfortunately, many 1H-NMR spectra are severely overlapped due to the multiplet structure caused by homo-nuclear scalar couplings. "Pure shift" NMR spectra, also known as broadband homonuclear decoupling, has been developed for disentangling overlapped proton spectra. The resulting spectra are considerably simplified as they consist of single lines, reminiscent of proton-decoupled 13C-spectra at natural abundance, with no muliplet structure. The number of signals typically decreases by almost an order of magnitude. Although selective and band-selective homonuclear decoupling of proton spectra have been used for a long time, it is not possible to achieve broadband decoupling of the whole spectrum by those methods and not in real-time, or to apply heteronuclear broadband decoupling sequences to the homonuclear case. Of the homonuclear broadband decoupling methods, there are 2D J-resolved methods, and methods based on constant time evolution methods, bilinear rotational decoupling, time-reversal, and slice-selective (Zanger-Sterk) decoupling. The Zanger-Sterk methods are the basis of the "Pure Shift" NMR experiment described and implemented below. There are based on frequency selective excitation during the application of a weak pulsed gradient. Thus, in effect, the desired 1-Dimensional broadband decoupled proton experiment is first acquired as a pseudo two-dimensional experiment and from these data the 1-Dimensional broadband decoupled proton experiment reconstructed.

An example is shown below (blue: conventional 1H spectrum; red: "Pure shift" NMR spectrum)



1. First acquire a normal proton spectrum. In particular, make a note of the 90 Degree pulse length that was obtained using the "pulsecal" command, or comparable means.

2. Make a note of the relevant sweep width and offset (centre of sweep width) that you would like for the "Pure Shift" NMR spectrum. Please note that the signal to noise is critically dependent on the sweep width, in a inverse relationship. That means, that the smaller the sweep width is, the higher the signal to noise ratio. The default is 4 ppm. For a 10 ppm sweep width, the acquisition times may have to be 4 times longer for a comparable signal to noise ratio.

3. Create a new dataset in experiment number n.

4. Read the "push1dzs" parameters in by typing "rpar push1dzs" or by selecting the parameters from the list in the /user subdirectory (set in "Source").

5. Insert the 90 Degree pulse length, sweep width into the corresponding "AcquPar" fields, or type "p1", "sw1". Set the receiver gain by "rga"<enter>; it will most likely be high as this experiment is intrinsically not very sensitive.

6. Start the data acquisition by typing "zg"<enter>.

7. The experiment with default parameters takes about 10 minutes. Increase the number of scans (NS) if the signal to noise ratio is unsatisfactory (best in multiples of 64).

8. Processing can be done while the experiment is running, but of course, not with the complete data set. The pseudo-2D-experiment can be processed with "xfb"<enter>. The reconstructed 1D "Pure shift" data set can be generated by typing "pshift"<enter>, and the results are written to the a new location (default is experiment number 1000+n). To process the dataset, use command "efp"<enter>, then "apk"<enter> like for a usual 1D experiment.


The standard parameters are necessarily a compromise, but should be fine for a start. For better data the parameters need to be modified but to avoid artifacts some rules need to be followed.

NS (number of scans) and DS (number of dummy scans) should ideally be multiple of the number of phase cycle permutations, 64. However, in practise, the distortions are slight, if values such as 1, 2, 4, 8, 16 are used.

SW2 (sweep width of proton dimension) should be as small as possible to keep the signal to noise level up. A value of over 10 ppm, such as 20 ppm, may produce spectra that have severe phasing issues.

SW1 (sweep width of duration of chunks) are typically 50-100 Hz. 60 Hz, or 0.1 ppm on a 600 MHz NMR instrument, may be a good value. 30 Hz may be indistinguishable from 60 Hz. Please note that the ratio of sw2/sw1 must be an integer to avoid artifacts. If the ratio is not an integer, the resulting "Pure shift" FID will contain discontinuities at regular (1/sw1) intervals, which will Fourier transform to give sidebands with spacing sw1.

TD1 (number of chunks) determines the limiting resolution of the final spectrum (sw1/td1), typically 8-32.

gpz2 (slice-selective gradient), typically 50-90%, determines how much of the spectrum is excited.

gpz2 (CTP gradient), typically 0.25-2%, a too small of a value may make some signals invisible.

cnst4 (number of extra data points to be discarded from the beginning of each chunk, to avoid artifacts caused by digital signal processor), typically 1-2.

d1 (relaxation delay), 1-5 * T1.


K. Zangger, H. Sterk, J. Magn. Reson. 1997, 124, 486-489.

J.A. Aguilar, S. Faulkner, M. Nilsson and G.A. Morris, Angew. Chem. Int. Ed.; 2010, 49, 3901-3903.

K. Zanger. Progress NMR Spectroscopy 2015, 86-87, 1-20.

The Manchester NMR Methodology Group.

Page created by Alexander Goroncy.

Contact Us

Chemistry Department

Department 3838

1000 E. University Avenue

Physical Science Bldg 204

Laramie, WY 82071

Phone: 307-766-4363

Fax: 307-766-2807

Email: chemistry@uwyo.edu

Find us on Instagram (Link opens a new window)Find us on Facebook (Link opens a new window)Find us on LinkedIn (Link opens a new window)Find us on YouTube (Link opens a new window)