University of Ottawa NMR Facility Web Site

Please feel free to make suggestions for future posts by emailing Glenn Facey.



Friday, August 17, 2012

Measurement of Long Range C H Coupling Constants

The stereochemistry of compounds is assigned very often with proton - proton NOE's by applying the 2D NOESY technique or the 1D selective gradient NOESY technique.  These methods fail, however when the distance between protons is too large to measure an NOE.  When faced with this situation, it may be possible to measure long range proton - carbon coupling constants which are able to provide the necessary information.  Three-bond carbon - proton couplings follow a Karplus relationship where the magnitude of the coupling constant is related to the dihedral angle between the carbon and the proton.  In some cases, these dihedral angles may be used to assign the stereochemistry.  Coupling constants are largest for dihedral angles of 0° and 180° and smallest for dihedral angles of 90°. The simplest way to measure the long range coupling constants is to collect a 13C NMR spectrum without 1H decoupling.  These spectra can be very complicated as can be seen from the figure below showing the C2 and C3 aromatic carbons of toluene.
Extracting specific long range carbon - proton coupling constants is quite tedious.  One way to simplify matters and obtain specific carbon - proton coupling constants is to apply the selective 2D heteronuclear J-resolved technique first introduced by Bax and Freeman in 1982 (JACS 104, 1099).  This method employs a 13C spin echo with a selective 1H 180° pulse applied simultaneously with the 13C nonselective 180° pulse. A version of this sequence is shown in the figure below with a shaped adiabatic 13C 180° pulse.

In this sequence one obtains a 2D spectrum with 13C in the F2 domain and the long range couplings to the selectively inverted proton in the F1 domain.  An example is shown in the figure below for toluene where the methyl protons were selectively inverted with a 20 msec Gaussian pulse.

All of the carbons coupled to the methyl protons are split into quartets in the F1 domain and the long range coupling constants which were very difficult to obtain from the coupled 13C spectrum can simply be read directly from the 2D spectrum.    

16 comments:

Craig Grimmer said...

I've been trying to find a way to measure long-range 1H-13C coupling on a 15 year old Varian Inova spectrometer with a 4NUC probe (without WFG) so this post has come at an opportune moment. The closest match I've got is "HET2DJ", without the selective or shaped pulses. I should be able to make limited shaped/selective pulses though PBox, even without a WFG - does anyone have a pulse sequence written with the selective and shaped pulses that they're willing to share?

Glenn Facey said...

Hi Ceaig,

The only shaped pulse you need is the 180 deg proton selective pulse. You can use a standard hard 180 deg pulse for carbon.

Glenn

Glenn Facey said...

Craig - sorry for the typo in your name - I have fat fingers.

Craig Grimmer said...

Hi Glenn.

No problem with the typo. Thanks for the post and for the follow-up. This is exactly what I've been looking for.

Craig.

Craig Grimmer said...

Here's an observation from some of my reading on this topic - at a time when almost every journal publication is accompanied by supplementary information of some kind, very few publications of new or improved pulse sequences/pulse programs include the actual program itself.

Fingolfin said...

@Craig

http://sermn.uab.cat/2012/03/measurement-of-the-magnitude-and-the-sign-of-small-njch-on-protonated-and-non-protonated-carbons/

Includes pulse program code for Bruker at the bottom of the post.

Craig Grimmer said...

Thank-you Fingolfin. For those interested in this topic, there's a mini-review here at http://onlinelibrary.wiley.com/doi/10.1002/cphc.201100748/abstract (ChemPhysChem, 2012, 13, 3, 645 - 660, Nath, Lokesh, Suryaprakash).

kubischkin said...

how about HMQC without 13C decoupling?

kubischkin said...

I mean, if you set the d2 500ms or like this, the INEPT transfer might be sufficient enough to see the long range in HMQC with the switched off decoupler. Actually, don't you have the J-coupled signals in HMBC already (it has by default no 13C decoupling)?

The other question: how can you extract the J(C-H) so precise (first digit after point) from a 13C-detected 2D? Normally, maximum you can go with a 1-2 second detected 13C spectrum is 1-0.5Hz.

Glenn Facey said...

Kubischkin,

Thank you for the question. An HMQC without 13C decoupling would not be the best way to measure long range 1H-13C coupling constants because HMQC's are generally acquired with short acquisition times (in t2) limiting the resolution in the 1H domain. Also they would not detect quaternary or carbonyl carbons (with the exception of aldehydes) which may show long range 13C-1H J coupling.

Glenn

Glenn Facey said...

Kubischkin,

You could use an HMBC to measure the long range couplings but you would have to make sure the acquisition time in t2 would be long enought to allow very high resolution in the 1H domain.

The couplings can be reported to +/- 0.1 Hz in this method because they are measured in the t1 domain not the t2 13C detection domain. The delay increment in the t1 domain is chosen to be small enough such that there is adequate resolution to precisely define the coupling constants.

Glenn

kubischkin said...

well, Glen, thank you very much for your comments!

Concerning the latter I don't think you got correctly what I meant. I understand that TECHNICALLY you read the J values from the indirect dimension F1. But... If you ever could have the resolution in the indirect dimension more than you have in the direct dimension?..

The observation resolution is defined by how long the collected FID last. Critical point is the one when the signal decays to the noise level. Normally you adjust the acquisition time to it, so you have the final resolution of 1/aq.

Carbon observation conventionally last 1 second. You can reach meaningful 2 seconds if you collect really long. But this is what 13C nuclei THEORETICALLY can give to you.

Now you tell, just because you have your 2D and you technically separated delta and J, you got the resolution of + - 0.1Hz. But the observed nuclei is still carbon. So my question was: could this even be possible? Without talking about the window function yet, what was the band width of the multiplets in F1?

Glenn Facey said...

Kubischkin,

Than you for the clarification. I think now I understand better what you were asking. I agree with you that the resolution in the F1 domain cannot be better than the best possible attainable resolution in the F2 domain. I also agree that the best "real observed resolution" (Hz per point) is determined by 1/AQ. I do not agree with your statement that "Carbon observation conventionally last 1 second". While it is common to use acquisition times of 1 second for 13C observation, the FID's in my experience are ususlly longer than 1 second if the magnet is shimmed well. I routinely use 1 second acquisition times, 1 second recycle delays and 30 degree pulses for 13C observations to save time. I also process the data with a 1 Hz exponential line broadening function to eliminate sinc wiggles due to FID trucation and improve the signal-to-noise ratio. Collecting data in this way optomizes the signal-to-noise ratio but does not produce spectra with the optimum attainable resolution. The theoretical line width at half height is determined by 1/(pi * T2) for each carbon. In reality, magnet inhomogeneity, decoupling efficiency and the presence of paramagnetic oxygen will make T2* < T2 and therefore ontribute to the line width. Even with these effects however, the FID's are typically > 2 seconds before decaying into the noise (one must have a high S/N ratio). It is possible to observe 13C lines with widths at half height of typically 0.3 Hz. In the BLOG post, the resolution in F1 is clearly better than 1 Hz (see the C4 resonance). Although I report the coupling constants to one place past the decimal, I do not claim that the error is +/- 0.1 Hz. I would estimate the error to be approximately 0.3 Hz.

Glenn

kubischkin said...

Now this is it! Thank you for the comment, Glen!

The picture was apparently too small for a not familiar viewer like me to look suspiciously through all given resonances.

Unfortunately, the only liquid spectrometer available to me now is the one with an inverse probe. Also, my field strength (600MHz) means quite quick CSA relaxation so important for carbon. Eventually, I barely get 1 second FID on this nuclei, regardless how good I shim.

Sure, on an adjusted machine you might get more. I'd agree with 0.3Hz maximal resolution. 1s is just the AQ from the standard Bruker experiment. But still... if you imagine chemists who will bring you their REAL samples: heavy molecules, dirty substances, low amounts, silica gel in NMR tubes etc - you'll probably end up with +/-1Hz in the best case ..)

IMHO...

ayhl said...

Hi Glenn,

I recently come across J-resolved experiments to look at long-range 1H-13C J-couplings. Is this 2D experiment very insensitive? A decent 1D 13C spectrum takes 4 scan but I am not seeing anything after 16 scans set in the 2D spectrum.


Thanks,
Andy

Glenn Facey said...

Andy,
I cannot seem to locate the original data from this post. If I recall correctly, I used a sample of 50:50 toluene:CDCl3. I don't recall the 2D data set requiring more than 1 - 2 hours. You may get a better idea of the sensitivity by taking a look at the original paper cited in the post.
Glenn