NMR terminology: "Transients"

I was reading a paper on hyphenating a gas chromatograph with an NMR. The authors describe a proof-of-concept type instrument. We have a couple of NMR experts here in SE, hoping to get some insight. The GC-NMR output is a plot of retention time (y-axis) vs. chemical shift (on the x-axis).

I wanted to confirm the meaning of the term "transient" as used in the figure caption of the following paper. The figure, taken from Analytical Chemistry (paper-behind paywall), shows how a GC is coupled to an NMR.

"Data acquisition parameters: 32 transients with 4k time domain points and a spectral width of 5618 Hz were accumulated with a relaxation delay of 500 ms."

1. Is transient just jargon for free induction decay signal? The complete term could be a transient signal/response. Also, if I understand correctly, 4k means that 4000 data points were collected in the FID after the NMR pulse, and the FID collection was delayed for 500 ms. I gather this "transient" wording is borrowed from electronics, where it refers to a quick change in the current in a circuit.

2. How does one control the spectral width of observation during data acquisition, say, as in this figure, it is 5168 Hz?

3. Acquisition time of 28 s (of FIDs or transients?). This time seems too long. Is this reasonable?

Article in question: Grynbaum, M. D.; Kreidler, D.; Rehbein, J.; Purea, A.; Schuler, P.; Schaal, W.; Czesla, H.; Webb, A.; Schurig, V.; Albert K. Hyphenation of Gas Chromatography to Microcoil $$\ce{^1H}$$ Nuclear Magnetic Resonance Spectroscopy in Anal. Chem. 2007, 79, 2708-2713; doi 10.1021/ac0617767.

• "Transient" for "decay" or whatever has a relatively short duration. Is a rather widespread term, in many techniques and fields. It can even be a pulse. Something Transient, in fact. Oct 22 '21 at 8:14
• Yes, but "transients" is a jargon of NMR. Of course, it has ten other meanings. Oct 22 '21 at 15:10

"Data acquisition parameters: 32 transients with 4k time domain points and a spectral width of 5618 Hz were accumulated with a relaxation delay of 500 ms."

"Transients" refers to the number of scans, i.e. number of times the experiment is repeated in order to obtain better signal-to-noise. (The SNR scales as the square root of the number of transients, or number of scans.) So in a sense, yes, it refers to the number of FIDs recorded. I don't know where the term comes from, though.

Also, if I understand correctly, 4k means that 4000 data points were collected in the FID after the NMR pulse

Yes, although they probably meant 4096. It's common in NMR to loosely use terms like "1k", "2k", "4k" with the understanding that it actually means a power of two. (The original idea was that $$2^n$$ points is "most compatible" with the FFT algorithm, although nowadays that's not a problem and if you really wanted exactly 4000 points it's perfectly fine.) If you type "4k" into TopSpin (Bruker's NMR software), it will set it to 4096 automatically.

For what it's worth, I don't like this shorthand (typing 4096 is just two characters more than 4k and is far more precise), but well.

and the FID collection was delayed for 500 ms

No, the relaxation delay refers to the time between experiment repetitions, i.e. the time between the end of the FID and the beginning of the first pulse of the next experiment.

How does one control the spectral width of observation during data acquisition, say, as in this figure, it is 5168 Hz?

The FID is a series of discretely sampled points. The sampling frequency determines the spectral width as described by the Nyquist theorem. For a spectral width of $$\pu{5168 Hz}$$, one would need to sample real data points at a frequency of one point per $$1/(2\cdot \pu{5168 Hz}) = \pu{96.75 µs}$$.

Acquisition time of 28 s (of FIDs or transients?). This time seems too long. Is this reasonable?

Well, in one single FID $$4096$$ points are sampled at a rate of one point per $$\pu{96.75 µs}$$, so the total acquisition time is $$4096 \cdot \pu{96.75 µs} = \pu{396 ms}$$.

Of course, the entire experiment duration will take longer than that. One single repetition is roughly $$\pu{900 ms}$$ long ($$\pu{400 ms}$$ acquisition plus relaxation delay of $$\pu{500 ms}$$), and this thing is repeated 32 times, so without any further information, it seems sensible to assume that the entire experiment takes around 30 seconds. Indeed, on p. 2712 of the article the authors write "Averaging 32 transients (over $$\pu{28 s}$$)...".

• Re: "I don't know where the term comes from, though" – I'm pretty sure it came about after the invention of Fourier NMR to distinguish collecting fids for a period of time, from averaging under a continuous wave experiment. The VnmrJ source code (now open source) has comments going back to 1986 or so, and some legacy hooks for CW experiments. You can see nt for "number of transients" clearly being used, even then (e.g. github.com/OpenVnmrJ/OpenVnmrJ/blob/…) Oct 28 '21 at 16:47
• @Landak, Thanks, this makes more sense. Transients in electronics has a specific meaning and now I am sure, it came from there. From OED, Electronics. A brief variation in current or voltage caused by a surge in power, the connection of a load, etc. Nov 1 '21 at 19:20
• Since they also use transient power in the code you linked...the name is obviously of electrical origin. Nov 1 '21 at 19:21
• Yes -- that's right. They're identical to [L]RC ringdown in electronics, but the link between electronics and NMR isn't very clear. I work in MRI – where "the bit of wire with capacitors in it wot collects your signal" is called an RF coil; in chemistry NMR it is a probe; to an electrical engineer it is a resonant (or "tank") circuit. Arguably it is a nearfield antenna – and the use of the word 'antenna' for NMR is itself incredibly controversial! Nov 2 '21 at 16:05
• @Landak, Yes, I had asked an editor of magnetic resonance journal editor... the use of radiofrequency "absorption" and emission is highly controversial. He said this is the myth spread by MRI people. Nov 3 '21 at 23:06