iNMR icon

Fourier Transform Options

Spectra acquired on different spectrometers with different pulse sequences require different kinds of processing. iNMR can guess which kind of processing is required by each spectrum. You can just trust it and ignore the details. If your spectrometer is old or rare, read this page.

The options have a clear meaning, therefore it's easy to set their values if you know the kind of spectrum you are processing or the kind of defect you need to reduce. You can also experiment by yourself what happens when you change a parameter or invert the status of an option.

use this option

to get this effect

Live !

The frequency domain spectrum is displayed before you click OK. Every time you change a parameter, the display is updated to show the effect on the spectrum. This mode is available for 1D spectra and for 1D extracts of multi-dimensional spectra.


The number of complex points in the final spectrum. This number can be reached by truncation, zero-filling or linear prediction.

Use Linear Prediction

Fast linear prediction is used instead of zero-filling. The parameters are set automatically by iNMR. To perform LP with parameters of your choice, choose Process > Linear Prediction before transforming the spectrum.

remove the first n points

Removes the first points of the FID. Useful when the true signal starts after an echo. Only used in solid NMR.

Swap Sides

Swaps the left and right sides. To make an example with words: “MOLFETTA” becomes “ETTAMOLF”. It's a reflexive operation. Numerically it consists in changing the sign of the even points of the complex FID.

Mirror Image

Inverts the frequencies into the spectrum. To make an example with words: “MOLFETTA” becomes “ATTEFLOM”. It's a reflexive operation. Numerically it consists in conjugating the FID (changing the sign of the imaginary component).

Real FT

Old Bruker spectra require this kind of FT. It is also necessary for some phase-sensitive 2D spectra, acquired with the TPPI protocol. In this case, use the normal FT along the direct dimension (f-2) and the real FT along the indirect dimension (f-1).


This is a convenience shortcut. It puts the spectrum in magnitude (= absolute value) representation after applying the rest of the processing. You can toggle between real and magnitude representations with the command Process > Magnitude.


Old analog spectra can suffer from a glitch in the center, if the two channels used for quadrature detection are not balanced. This option subtracts form each channel the average of the last 6% (tail of the FID), thus alleviating the problem.


When processing the indirect dimension of a phase-sensitive 2-D spectrum, the imaginary part is normally discarded. At the end of processing, it is no more possible to correct the phase along the direct dimension. With the HyperComplex option you don't discard the imaginary part and you can still correct the phase along both dimensions, at the cost of using a double amount of memory. The value of this option is not stored when you save a document.


This operation is the pillar of multidimensional phase-sensitive acquisition. It is only required when processing along the indirect dimension. Traditional phase-sensitive spectra require the phase-sensitive shuffling. Gradient enhanced spectra (Rance-Kay), like the phase-sensitive HSQC, require the echo-antiecho shuffling. In the case of 3-D Varian spectra, the Rance-Kay editing is already applied when the data points are read from disk, so you should select the phase-sensitive shuffling.

Apply these Weights

The main switch for weighting. You can use it to force a live update (there is no live update when you type the value of a weighting function). It is safe to leave this switch always on. You can simultaneously apply the functions here below, in any combination.


Multiplies the FID by the exponential function:
f(t) = exp( - λ t ).
Whose frequency domain equivalent is a Lorentzian curve:
F(ω) = λ / (λ2 + ω2).
The width W of the Lorentzian curve is given by the relation:
λ = π W.
From a practical point of view, W, measured in Hz, is the parameter that best describes the applied weight, and it's also the parameter requested by iNMR. Line broadening is accompanied by sensitivity enhancement. You can specify a negative value for W, that corresponds to a positive exponential and causes a resolution enhancement.


Multiplies the FID by the Gaussian function:
f(t) = exp( - σ2 t2 / 2 ).
Whose frequency domain equivalent is a Gaussian bell:
F(ω) = √(2 π) exp[-ω2 / (2 σ2 )] / σ.
The width W of the bell is given by the relation:
σ = π W / √ loge2 = 1.2 π W.
Like in the case of the exponential, the parameter requested by iNMR is the linewidth W, in Hz. In this case, however, it can only have positive values. The gaussian is an apodization function, which ensures that the FID decays to zero and thus removes the wiggles that would otherwise appear in frequency domain.

Sine Bell

Shifted Sine Bell. When the shift is 90°, it becomes a cosine bell. In all cases, it goes to zero at the right extreme (180° shift).

Squared Sine

The square of the function above.

Cut After ... %

Works as an additional parameters of the above sine-bell functions. It can go from 5% to 200% of the FID and becomes the effective length of the sine bell (where the sine becomes zero). When it is less than 100% the FID is truncated. If you already truncate the FID with the size menu, there is no need of setting this parameter. It makes sense if you want to combine truncation with zero-filling. When the value is more than 100%, the bell is truncated (it would go to zero too late, beyond the end of the FID).

Can be used without the sine-bell functions and gets completely different meanings. For values under 100%, it applies a trapezoidal shape. For values between 100 and 200, it applies a linear ramp. For higher values, it applies a Traficante function optimized for (value-200) Hz.

Multiply 1st Point By

This can be useful to correct the baseline. It applies a user-defined weight to the first point only of the complex FID. Use this correction when, for whichever reason, you don't like to correct the baseline after the FT. The value 0.5 is routinely used in many 3-D experiments.


This command skips processing altogether. Data points are subsequently treated as a frequency domain spectrum. All the operations that are not allowed with a FID, like zooming, panning and phasing, become possible. You use faking when you want to inspect the FID closely.

If you hold down the Option key, the command Process > Fourier Transform becomes FT now, which skips the dialog and performs the processing with the parameters already stored (if old parameters are not found, a new set is created with default values).

The gears icon inside the palette (or its menu equivalent Run) are more powerful than the command “FT now”. If iNMR finds the parameters for peak suppression, it performs that as the first operation. If the parameters for linear prediction are found, that is performed in sequence. Finally, FT is performed.

Related Topics

About Processing