Dear José, if you have an analysed spectrum (parameters: chemical shift and coupling constants), it is straightforward to calculate the spectrum at any desired magnetic field. There are several software packages offering this possibility, for example the "NmrSim" module in Bruker Topspin.

If you only have the spectrum (no analysis and no extracted parameters), you will probably fail.

Hello Wolfgang Baumann, thank you very much for your reply.

In my case, we would not have analysed parameters. We would be interested in transforming the spectrum of real samples (mixture of complex signals) to see if the result is predictable on a low resolution desktop. We have seen software that allows us to simulate with an imput of parameters, but it would not be what we need.

I don't know if it would be possible to apply smoothing to the FID signal to make it "lose" resolution or something!

Hi José Raúl Belmonte-Sánchez

It is unfortunately not possible to lower the resolution of a spectrum in the way you wish.

The resonance frequencies you see in the spectrum are what they are. If you had, for example, a doublet at 1 ppm, with a coupling constant of 10 Hz ad recorded it on a 500 MHz spectrometer, you would see one peak at 495 Hz and one peak at 505 Hz, or 1.01 ppm and 0.99 ppm. The fid does not contain information about how these peaks are connected, and a spectrum containing two singlets at the same positions would give an identical fid.

Recording the same sample on a 600 MHz instrument would give a doublet with peaks at 595 and 605 Hz, or 1.0083 ppm and 0.9917 ppm for the case of a doublet and if it were two singlets instead the chemical shifts would be the same as for the 500 MHz instrument.

The only way to know whether you are dealing with a doublet or two singlets is to analyze the rest of the spectrum and determine the chemical shifts and coupling constants.

Finally, higher order effects are more pronounced at lower fields, so the shapes of the signals themselves will be different, and this will again require the knowledge of coupling constants.

Dear Jose, there is no way around of what has been said above....for a very hand-wavy estimated I'd use a lb of 8 (or so) and do a em+ft+pk sequence...but as said this is REALLY very hand-wavy! A line-width of 2 Hz (which s 2/400 ppm on 400 MHz) will look at 60 MHz like 2/60 ppm...this works well for singulets, but not for multiplets! If you have complex mixtures this might give you a visual impression....

Alfred

Dear Jose,

I also agree to the statements above. Alfred Ross described the fastest way to decrease the resolution of your spectrum.

The FID is multiplied (usually) using a so-called window-function. You can use an exponential window function "em" like Alfred Ross described. The line-broadening (lb) is similar to the exponential decay of that function. If you are using a large value here, is makes your peaks broader.

If you are using lb of 8 Hz, for instance, this adds 8 Hz to the half width of your peaks.

If you are using topspin, go to procpars in your dataset and change LB [Hz] accordingly. followed by reprocessing typing "efp" in the commandline.

Since resolution is proportional to "line with/spectral width" you can display the wanted resolution in that way.

As Alfred Ross said, the situation for the resolution within multiplets is different since coupling constancts act not according to over all peak positions (nor changing with the field in terms of J-coupling).

Off course what you get out is not the same as if you used an 60 MHz spectrometer. If you go down in field, the peak width in Hz stays constant while the spectral with over which your resonances are distributed shrinks. Or in other words all positions of the peaks (resonance frequencies, or chemical shifts) are changing. The resolution shrinks because the same number of peaks (with a similar width) are distributed at less "space". What you are doing with lb (line broadening) is that you, by contrast, increase the width of the peaks while the spectral width (spectral "space") stays constant.

Off course the spectrum what you will get out has a differnent ppm scale as a 60 MHz spectrum would have. Among other differences.

Many Greetings,

Kristof

I think your purpose using low NMR to predict the mixture signals is not realistic work.

First of all, thank you all very much for your contributions. I understand what you are telling me and I see that the simulation based on an FID obtained is a much more complex case than I thought. My idea was based on the idea that, given a FID obtained "without manipulation", it was possible to apply some kind of smoothing, point decrease or mathematical algorithm that would worsen its "quality". As the high-resolution FID contains all the information contained in the FID of a low-resolution device, I found it conceptually feasible to carry out, even though I did not have the necessary knowledge. I would then apply a FT to obtain the resulting spectrum.

I have learned a lot from your comments and, for now, by applying a Gaussian lb I have obtained the most similar spectra despite the differences you explain about the coupling constants. It is not ideal, but it gives me a superficial idea of the target.

Of course, this is even more complex. The decrease in field will impact the appearance of the signals in more than linewidth and signal to noise. You will also see more strongly coupled signals because the chemical shift separation (in Hz) is less than a high field instrument. The attached simulation shows what you might expect at 60MHz and 500MHz.