Measuring of water diffusivity with NMR in building materials is hindered by the presence of internal magnetic fields originating from the magnetic impurities (Fe). To investigate the diffusion of water and ions in building materials, a stimulated echo NMR technique is applied. A new analytical equation for the long time decay in the presence of spatially varying internal field gradients is derived. This equation is experimentally confirmed and applied to stimulated echo decay in the materials with high internal gradients (sintered crushed glass and fired-clay brick) and a standard material without internal gradients (glass filter). In figure 1 the measured stimulated spin-echo as a function of interpulse time and for various gradients strength for fired-clay brick.

Figure 1. Ratio between the stimulated echo and primary echo in fired clay brick plotted against the spin-echo time.
 a) interpulse time t is varied at constant gradient strength G = 188 mT/m.
b) G is varied at t = 410 ms. The solid lines reflects the decay due to longitudinal relaxation only.
 Dashed lines are fits of a mono-exponential decay.

The diffusivity is determined from the slopes of the mono-exponential decay in the long time limit. The diffusivity is constant and limited by the tortuosity of the pore structure. Tortuosities of different samples are calculated, showing the excellent agreement with the macroscopic tortuosities measured by electrochemical impedance spectroscopy method (see table 1).
 
 
 
Glass filter
Sintered crushed glass
Fired-clay brick
T1 (s) (SE)
1.50± 0.02
0.36± 0.02
1.04± 0.02
T1 (s) (SR)
1.49± 0.01
0.37± 0.01
1.10± 0.02
D (10-9 m2/s)
1.20± 0.05
1.55± 0.05
0.41± 0.02
alpha (SE)
1.8± 0.1
1.5± 0.1
5.6± 0.5
alpha (EL)
-
1.4± 0.1
5.9± 0.2

Table 1. Comparison of the stimulated echo (SE) results with the values of T1 obtained by saturation recovery method (SR)
 and the electrolytically (EL) determined tortuosity alpha . The thickness of the glass filter sample was insufficient to perform electrochemical tortuosity measurement.