Monday, December 5, 2016

ASAP - Acceleration by Sharing Adjacent Polarization

NMR is a non destructive technique and so reducing acquisition time can be used either to increase sample throughput, or to lower the detection limit. To lower the detection limit a given amount of time is used to acquire a greater number of scans than would normally be acquired. A relatively recent method for reducing acquisition time is ASAP (Acceleration by Sharing Adjacent Polarization)1. The ASAP technique speeds acquisition by greatly reducing the length of the relaxation delay.

All NMR experiments include a relaxation delay, a period in which the magnetization is allowed to relax back to thermal equilibrium. The relaxation delay is executed before every scan. In nearly all cases, a relaxation delay of one second is used in the Facility's standard parameters.

The figure below shows the pulse sequence diagram for a standard HMBC in blue at the top, and for an ASAP-HMBC in red at the bottom. Normally these pulses sequence diagrams are just used to indicate timing and are not drawn to scale, but in this case the horizontal axis has been drawn to scale and indicates time in seconds. In the standard HMBC the relaxation delay of one second takes about 23 of the time taken to record a scan (1.35 s). In the ASAP-HMBC a short 25 ms TOCSY spinlock period (the solid red block), is introduced and the relaxation delay is much shorter (30 ms) reducing the total acquisition time (0.40 s) to about a third of the standard HMBC.

In a heteronuclear experiment on a natural isotopic abundance sample, signal is only detected from protons attached to 13C. The protons attached to 12C do not contribute to the signal and remain at thermal equilibrium. The spinlock period in the ASAP-HMBC mixes the polarization of all the protons, effectively transferring equilibrium magnetization from 12C-attached protons to the 13C-attached protons and allowing the relaxation delay to be greatly reduced.

The ASAP technique can only be applied to heteronuclear experiments on samples at natural isotopic abundance but ASAP-HSQC2 and ASAP-HMBC3 experiments have already been published. The main drawback of ASAP experiments is that the acquisition period now takes up the majority of the time to record a scan. For experiments that use composite pulse decoupling during acquisition (like HSQC and H2BC) this leads to heating. A recent attempt4 to minimize this problem used chunked data acquisition which enabled the decoupling to be turned off for parts of the acquisition. This reduced heating but complicated data processing.

Overall, ASAP is a simple technique that offers significant reductions in acquisition time. It can also be used in conjunction with non uniform sampling and the combination of these two techniques should yield much faster experiments and decreased detection limits.

1. Erics Kupce, Ray Freeman
"Fast multidimensional NMR by polarization sharing."
Magn Reson Chem. 2007 45(1):2-4.

2. David Schulze-Sünninghausen, Johanna  Becker, Burkhard Luy
"Rapid Heteronuclear Single Quantum Correlation NMR Spectra at Natural Abundance"
J. Am. Chem. Soc. 2014 136(4):1242-1245

3. Julien Furrer
"A robust, sensitive, and versatile HMBC experiment for rapid structure elucidation by NMR: IMPACT-HMBC"
Chem. Commun. 2010 46(19):3396-3398

4. Ikenna E. Ndukwe, Alexandra Shchukina, Krzysztof Kazimierczuk, Carlos Cobas, Craig P. Butts
"EXtended ACquisition Time (EXACT) NMR-A Case for 'Burst' Non-Uniform Sampling"
Chemphyschem 2016 17(18):2799-2803


  1. After the acquisition period, if the magnetization still has not gone back to the equilibrium position, does it still generate an FID? If so, for conventional HMBC, why not just increase the acquisition time by 1s and completely remove the relaxation delay? This seems like it should give the same experiment time but maybe better signals.

    1. If the magnetisation has not returned to equilibrium after acquisition and the relaxation delay on the next scan it will contribute to the signal but it will not give maximum signal. If you imagine the magnetisation is 30o from the +z-axis, applying a 90o pulse will push it beyond the xy-plane. Resolving that magnetisation into two vectors, one along the -z-axis and the other in the xy plane will give you signal in the xy plane, but it will be smaller than what you would have got had you started aligned along the +z-axis. Also, the portion of the magnetisation that resolves to the -z axis will experience the pulses in the rest of the pulse sequence and may end up getting moved to a position where it creates detectable signal. This is a particular problem in COSY experiments where a short relaxation delay leads to artifacts that look like an extra, steeper diagonal that runs from bottom left, to top middle, then bottom right. In this case strong gradients can be used to remove the unwanted magnetisation and thus the artifacts.

      Increasing the acquisition time and reducing the relaxation delay is definitely an option. The drawback of increasing the acquisition time is that after a certain point (1.2 x T2) you no longer increase the signal, you just collect more noise. I typically set the acquisition in 2Ds to acquire 4k points because this keeps the processed data to an easily manageable size, but as computers have become more powerful this is less of a limitation.

  2. For the ASAP-HMBC, will it reduce S/N if the proton is isolated and not coupled to any other proton?

    1. This is a good point that I had not thought of. When I compared a standard HMBC to an ASAP-HMBC there were differences in peak intensities which I thought might be due to different T1s, but differences in ability to transfer polarization may be responsible as well. I will have to look at my comparison spectra again to see if that makes sense.

  3. For ASAP-HMBC, if a proton is adjacent to other protons but with very small coupling constants, will the TOCSY spin-lock efficiently help the recovery of its magnetization? I’m asking since I vaguely remember TOCSY is able to see correlations with very small couplings better than COSY.

    1. I suspect that the size of the coupling doesn't matter very much. If you look at the "Polarization Sharing" section in the Kupce and Freeman paper referenced in the post they say the TOCSY mixing is "mediated by the scalar coupling", but "is a superposition of several sinusoids." I think that as long as the period is short enough to avoid presaturation you'll be OK. Kupce and Freeman and Furrer recommended a spinlock of 40ms, while Ndukwe et al used 25ms. I have used 25ms.