Abstract
The standard prepolarized and nonpolarized measurement sequencies used in fastfieldcycling (FFC) NMR relaxometry [1] suffer from two major drawbacks:
(a) Due to the magnetization evolution during switching intervals, the longitudinal magnetization evolves with increasing τ towards nonzero values. Consequently, FFCNMR decay curves must be fitted with an adjustable 'offset' parameter, in addition to the decay rates and weights of individual sample components (for example, three parameters in the case of a monoexponential fit).
(b) Measurements are carried out using different sequences at high relaxation fields (NP) and at low relaxation fields (the crossover field being approximately half of the polarization field). For reasons which are not yet quite clear, this sometimes leads to a statistically significant discrepancy between the two 'sections' of the resulting NMRD profile.
We propose new measurement sequences consisting in a combination of RF pulsephase cycling, receiver cycling and magnetic field cycling, such that all switchingtime interval effects get cancelled, while the sample prepolarization effects and the subsequent decay are enhanced.
In such sequences, the acquired signal S(τ) decays rigorously to zero when τ grows to infinity (which is why we call them nullbiased). Essentially all classical FFCNMR sequences can be cast in this way (the basic NP/PP, IR, IR_CPMG, ...), some of them in a way which does not at all reduce the efficiency of the measurement (i.e., the total signal range which can be achieved in a given time).
The nullbiased sequences offer several important advantages:
(1) Since S(∞) is strictly null and does not have to be estimated experimentally, there is no need to acquire portions of the decay curve where the signal is already almost stable (this leads to a time saving).
(2) For the same reason, the data can be fitted without the adjustable offset parameter (for example, a twoparameter fit is sufficient in the case of a monoexponential decay). Considering the effect of number of adjustable parameters on their confidance intervals, this fact alone improves the precision of the estimated T_{1} by a large factor (~10).
(3) In most cases, the same sequence can be used to acquire the NMRD profile over the complete range of field values (for example, from 5 kHz to 40 MHz). As a result, the measured NMRD profiles are internally more coherent and no discrepancy between highfield data and lowfield data can occur.
We discuss also an apparent drawback of the new sequences consisting in the fact that they enhance the visual impact of random field variations between consecutive scans. Analysis shows, however, that traditional measurements are burdened by the same instabilities, even though they are to a large extent masked by the switchinginterval signal components.
References:
 G.Ferrante, S.Sykora, Technical Aspects Of Fast Field Cycling,
in Adv.Inorg.Chem., Eds. R.van Eldik,I.Bertini, 2005, 57, 405470.
 C.Radhakrishna Rao, Linear Statistical Inference and its Applications, John Wiley & Sons, 1973.
