water signal the 1:1:1 triplet of the NH₄ signal should not be observed in the normal ¹H NMR spectrum.

Tune the probe-head carefully to the rather polar solution; this may be a bit difficult in some cases. Since the sample contains no lock signal, turn the lock and the field sweep off. Obtain a normal ¹H NMR spectrum of the sample, adjust the spectral width and the offset, and redetermine the 90° pulse on the water signal (see Exp. 2.7) under these conditions. Load a 1D version of the CPMG (Carr–Purcell–Meiboom–Gill) pulse sequence. You have to set:

td	32 k
sw	10 ppm
o1	on ¹H resonance of water signal
p1	90° ¹H transmitter pulse [14 μs, 0 dB]
p2	180° ¹H transmitter pulse [28 μs, 0 dB]
d1	1 s
d2	300 μs
n-value	(loop parameter for CPMG) 200, leading to a delay of 0.12 sec between the first 90° pulse and start of the acquisition
ns	1

5. Processing

Use standard 1D processing for ¹H NMR spectra (see Exp. 3.1).

6. Result

The figure shows at the top the normal ¹H NMR spectrum of the solution obtained on a DRX-400 spectrometer using an inverse probe-head without spinning. Below is the result of the CPMG sequence, showing virtually no water signal and revealing the signal of an anomeric proton, which was completely covered by the water resonance.

7. Comments

The exchange between NH₄ and water protons is pH-dependent and rather slow at low pH. If one takes Coca Cola directly out of the bottle and adds NH₄Cl, one observes a sharp 1:1:1 triplet of the ammonium ion at about δ_H = 7 due to the spin–spin coupling to ¹⁴N. At about pH = 6.5 the exchange is very effective, leading to a shortening of the water T₂ by a factor exceeding 100. Thus, in the spin-echo sequence a total delay value can be found such that the water signal will not be recovered, whereas the signals of non-exchanging protons are mainly unaffected. The optimum delay may be field-dependent. As well as NH₄Cl, other exchange reagents have been reported, such as guanidinium chloride for working in a different pH range.