Thursday, November 18, 2010

The Phase of an NMR Spectrum

Most students know that the phase of an NMR spectrum has to do with the degree to which the NMR resonances are above or below the baseline of the spectrum (i.e. the amount of absorption and dispersion character). Most students have also learned that the phase of a periodic time domain function depends only on the value of the function at time zero. Thus, the only difference between a cosine and a sine function is where the function starts at time zero. A sine is said to be 90° out of phase with respect to a cosine. Many students do not understand the connection between the phase of their NMR spectrum and the phase of the periodic time domain function giving rise to their spectrum by way of a Fourier transform. The figure below is an attempt to make the connection.
The left-hand portion of the figure shows equilibrium magnetization vectors being rotated by radio frequency pulses. 90° pulses along the x, y, -x and -y axes rotate the z magnetization vector to the -y, x, y and -x axes of the rotating frame, respectively according to the right-hand screw rule. After the pulse, the magnetization vector rotates in the rotating frame of reference at a frequency equal to the difference between the transmitter frequency and the frequency of the NMR resonance. In the figure, the magnetization is assumed to be rotating anti-clockwise representing an NMR resonance with a positive frequency with respect to the transmitter frequency. The NMR spectrometer measures the time dependent voltages on two of the four orthogonal axes of the rotating frame separated by 90° (quadrature detection). The time dependent voltages are proportional to the amount of magnetization on the axis as a function of time. One of the two time dependent voltages is called the "real" signal and the other is called the "imaginary" signal. Together these two signals make up the complex free induction decay (FID) which is Fourier transformed to produce the NMR spectrum. In the figure, the -y(t) voltage is the real FID and the x(t) voltage is the imaginary FID. The figure shows representations of the real and imaginary FIDs after the delivery of pulses along each of the orthogonal axes. Note that the phases of the real and imaginary FIDs depend on the pulse delivered. For example, after a 90°x pulse, the magnetization resides on the -y axis. The real FID (along the -y axis) starts at a maximum (cosine) and the imaginary FID (along the x axis) starts at zero (sine) and increases as the magnetization vector rotates anti-clockwise. The Fourier transform of the complex FID produces a real spectrum (typically the one displayed to the user) and an imaginary spectrum (typically not displayed to the user). Note that the degree of absorption vs. dispersion character in the spectrum depends on the phase of the FID signals. If an NMR spectrum is not in phase, perhaps due to a receiver dead time problem, it can be corrected after the collection of the data by calculating the phase angles needed to put the real spectrum enitirely in absorption mode and the imaginary spectrum entirely in dispersion mode.

3 comments:

Glenn Facey said...

The figure in this post was corrected on March 17, 2016 after an error was pointed out by Huldrych Egli. Thank you Huldrych.

Anonymous said...

My question has to do with the receiver phase. One usually sets the receiver phase such that most of the signal (when on resonance) is either on the real or the imaginary parts of the fid. In our spectrometer I noticed that this receiver phase is not stable over time, one needs to modify it occasionally to keep the fid in the right phase. Is this a magnet, probe, or spectrometer issue? I have also noticed that the probe wobb tends to drift slightly over time.

Thank you.

Glenn Facey said...

Anonymous,
The drift of the receiver phase is most likely a spectrometer issue. Note however that it is not necessary to have all of the on-resonance signal in either the real or imaginary channel. The phase of the spectrum can always be corrected after the data collection.

I have also noticed drift in probe tuning. I have always attributed it to temperature fluctuations, as it is often much more pronounced while running pulse sequences with higher duty cycles producing heat.

Glenn