Someone had asked me about it before and I investigated a bit, see attached the pulse sequence I coded. There is a reason why you have not heard of it, it is not generally a better option than a carbon run with appropriate parameters. If you check the papers of both DEFT and UDEFT, they are compared with a carbon run with a 90 pulse, but carbons are normally run with a 30 pulse. A 90 pulse would be optimum if the relaxation delay was infinite:

https://mriquestions.com/optimal-flip-angle.html

And we get better SNR per time using lower flip angles:

https://nmr.chem.ucsb.edu/protocols/fast1pulse.html

In addition, for DEFT to work well, you need to have a short acquisition time, otherwise magnetization relaxes and the pulse immediately after acquisition puts the magnetization back in the transverse plane and hence leading to signal loss, rather than flipping it to the z axis. So you have to sacrifice resolution in order to gain sensitivity, unless the T2s are extremely short and T1s extremely long.

So in my experience, with short acquisition times, DEFT gives better results than a 90 Carbon, but worse than a 30 Carbon.

The main reason to do a 90 Carbon is for quantitation, but DEFT does not help with that.

If you want to save time extracting carbon information, you could run an HMBC with a large number of increments.

https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/pdf/10.1002/mrc.1884

Best regards

Adolfo