T1 Noise

Reducing Apparent T1-noise in your processed 2D data.

User often ask how to reduce apparent T1 noise in their 2D data. Many aspects of 2D processing can contribute to what may appear to be T1 noise. The first step is to try to determine as much as possible what is creating the problem. To do this you need to look at the transformed data after the D1 transform and before the D2 transform. Try loading some sample columns and saving these to files. Try processing these and look for anything that may produce a T1 ridge in the transformed data. Baseline, phasing and appodization problems can all cause these sorts of problems.

Anything that causes a distortion in the baseline can produce artifacts that mimic T1 noise. Baseline artifacts cannot always be removed entirely by baseline correction. Therefore it is best to remove anything that contributes to baseline distortion as early in the processing cycle as possible. Many spectrometers tend to mess up the acquisition of the first few points. Since the first few points are critical for the baseline anything that can be done to correct these points will generally improve the baseline and hence the look of the entire spectrum. For this reason many people tend to use the lpf command to correct the intensity of the first points in D1 and sometimes D2. Linear prediction is used to estimate the value of the first few incorrectly acquired points using the values of subsequent points. Some people find that muliplying the first point by 0.5 works very well.

Often errors in phasing can mimic T1 noise. It is therefore critical that the phase be determined as accurately as possible. The rephase macro can be used to correct the phase of an already processed 2D spectrum very easily.

Apodization is also very important as it determines the extent to which the tails of peaks extend from the center. Felix provides many types of apodization functions including sinebell, sinebell squared, skewed sinebell, skewed sinebell squared, exponential linebroadening, gaussian linebroadening, kaiser and trapezoidal functions. Try various apodization functions until the data looks good and there is not much tailing of peaks. Remember that if you process your data as States then you would normally apodize over a number of points equal to one-half the total number of acquired fids. This is because the data is complex in D2 and a complex vector contains half as many complex data points as the corresponding real vector did. Incorrect apodization in D2 is probably the most common cause of ridges along D2.

After you have decided on the appropriate processing parameters for the second dimension then go ahead and process D2. Then take a look at the phasing and baseline for the columns. If you need to rephase or correct the baseline in the D2 dimension you can do this after the dimension has been processed by using the appropriate commands under the Process pulldown.