Experimental procedure
Once the mineral pastes were placed in their respective test tubes, at least two
days were allowed for moisture equilibration before the unfrozen water content
was determined. Following this period, the sealed tubes containing minerals and
solutions were placed in a bath preset to an initial test temperature of about 20C. If
the analysis temperatures were within the range of 20C to 20C, the bath heat-
transfer fluid was an ethylene glycolwater mixture. If the analysis required tem-
peratures below 20C, a second bath that used methyl alcohol as a heat-transfer
fluid and was capable of achieving temperatures of 70C was utilized.
Following a temperature equilibration time of approximately 1 hour, the sample
tubes were sequentially removed from the bath, wiped dry, and inserted into the
nuclear magnetic resonance (NMR) probe for analysis. (The mean background sig-
nal intensity due to the test tube was measured by placing an empty test tube
in the NMR probe.) The peak intensity for each temperature was recorded. The
NMR probe used for these analyses was a Praxis model PR-103 analyzer. It was oper-
ated in the 90 mode with a 0.2-s clock and at a fast scan speed. The first pulse ampli-
tude in the 90 mode was measured for each sample, starting with the first test tem-
perature of about 20C. After about 4 s (the time required to measure the NMR signal
amplitude) the samples were reinserted into the bath. When all the samples were
analyzed, the bath temperature was lowered by about 3C and allowed to re-equili-
brate at this new test temperature. This process was repeated until a temperature of
about 0C was reached. (All of the above-0C NMR measurements were used to de-
termine the paramagnetic effect or the increase in signal intensity with decreasing
temperature. This relationship is discussed by Tice et al. [1981, 1982].)
Once measurements above 0C were completed, the constant-temperature bath
was cooled to about 5C to initiate ice formation in the samples. The constant-
temperature bath was then raised to 0.5C, and the samples were allowed to equili-
brate at this temperature for several hours. After equilibration, the peak intensity and
temperature were recorded. The temperature of the constant-temperature bath was
then lowered in small increments. The signal intensity and temperature were recorded
after the sample was allowed to equilibrate at each temperature increment. The samples
were cooled until the measured signal intensity matched that for an empty test tube--
indicating that no liquid water remained in the sample (Tice 1982).
Following the last NMR cooling measurements, which in some instances were
taken at temperatures as cold as 77C, the warming run was begun. An analysis
interval of about 4C was selected at the lower temperatures, at which small changes
in temperature have little effect on signal intensity. At temperatures near melting,
intervals between equilibration temperatures were reduced because liquid-water
contents were increasing more with temperature. At around 1C, the interval by
which the temperature was changed was reduced to about 0.1C. This approach
provided good coverage within the region where the unfrozen watertemperature
relationship is so critical.
Unfrozen water contents were calculated by first regressing the above-0C NMR
readings minus the background against their respective temperatures for each
sample and extending the resultant line to the lowest temperature where experi-
mental data were obtained. Projected first-pulse amplitudes were calculated for
each experimental temperature. The gravimetric water content for each sample had
previously been determined by oven-drying at 110C for two days. A ratio between
the sample water content and the projected first-pulse amplitude was developed;
unfrozen water contents were calculated by multiplying these first-pulse ampli-
tudes by their respective ratios to obtain a value for each temperature (Tice et al.
1981, 1982).
6