The Temperature Dependence of Water Activity in Foods
Two models of AquaLab water activity meter, the CX2 and CX2T
THE CX2T VERSION IS equipped with a special sample chamber which can be connected to a constant temperature water bath. Water activity measurements can therefore be made at a fixed, constant temperature. The "T" version is more expensive than the standard version, and the constant temperature bath is an additional expense. Questions therefore arise about the importance of sample temperature control for accurate water activity measurement. We present here some results of our research on this matter.
There are two reasons for sample temperature control: to attain temperature equilibrium between the sample and the humidity sensor, and to compare water activities of different samples independent of any effects temperature may have on their water activity.
Many instruments control the sample temperature to assure that the sample and water activity sensor are at the same temperature. If water activity is sensed using a relative humidity sensor, and there is a 1° C temperature difference between the sample surface and the sensor, the humidity read by the sensor can be in error by as much as 0.06. For acceptable accuracy this type of instrument must therefore be designed to assure that the sensor and sample temperature are within about 0.05° C of each other.
Temperature and accuracy.
AquaLab measures the sample temperature, and determines the vapor pressure in the head space using a cooled mirror dew point sensor. A built in microcontroller computes the water activity from these measurements. Differences between sample and sensor temperature therefore have very little effect on AquaLab's accuracy.
Why constant temperature?
Decagon supplies a version of AquaLab with temperature control because the water activity of most food samples changes with temperature. The unsaturated salt solutions that Decagon uses for calibration standards maintain constant water activities over the normal temperature range of food samples, but most other samples change water activity to some degree as temperature changes. Even the saturated salt standards described by Greenspan (1977) change with temperature because the solubility of the salt depends on temperature.
Unpredictability of foods.
Labuza (1984) provided a thermodynamic model for estimating the temperature effect on water activity in foods, but the parameters in the model must be empirically determined. In the examples given by Labuza, water activity increases as temperature of the sample increases, but the water activity of most saturated salts decreases as temperature increases. One can therefore not predict even the direction of the change of water activity with temperature since it depends on how temperature affects the factors that control the water activity in the food.
Temperature and the matrix.
Solubility of solutes can sometimes be a controlling factor, but control is usually from the state of the matrix or food polymer to which the water is bound. Since temperature can have a profound effect on the state of that matrix, (glassy vs rubbery state) one should not be surprised that temperature affects the water activity of the food. Labuza showed that the effect of temperature on water activity is negligible in high moisture foods, but in intermediate and low moisture foods a 10° C change in temperature can result in a few percent change in aw.
Factory tests.
To provide guidance for customers who need to decide which instrument to buy, we purchased samples of six intermediate and low moisture foods and used an AquaLab CX2T to determine the water activity at temperatures of 10, 20, 30 and 40° C. The samples used were chocolate peanut butter cups, granulated dog food, frosted coconut cake snack, bakery chocolate chip cookies, sweetened toasted oat cereal, and powered, dried soup mix.
Experiment results.
Samples were sealed in Decagon's plastic sample cups, and the same set of samples was used for all temperatures. Samples were run in triplicate, and all measurements were completed within one working day to assure that samples hadn't lost water during the experiment. The measurement sequence was from cold to hot. Some measurements were repeated at 10oC to confirm that sample water activity had not changed during the experiment. Single samples of beef jerky, chocolate syrup, and sausage-all relatively high moisture foods-were also measured.
Figure 1 shows the results of the experiment. The foods clearly show a wide range of behavior. Some increase water activity with temperature, while others decrease. Many hardly change, but the soup mix changes substantially between 10 and 30° C. The average standard error of the mean for the three samples was 0.003, or about the size of the dots on the graph. So most of the changes shown on the graph are statistically significant. The errors for the peanut butter cups, however, were about 3 times the average. The beef jerky, chocolate syrup, and sausage showed negligible change in aw with temperature, and are therefore not included in the figure.
Figure 1. Temperature dependence of water activity in six intermediate and low moisture.
To determine whether you need a CX2T or a CX2 it helps to look at the data from Fig. 1 in a different way. The slopes of the lines in fig. 1 are plotted as a function of temperature in Figure 2. You can see the change in water activity with temperature is typically less than 0.002 per degree.
Figure 2. Change in water activity per degree change in sample temperature as a function of sample temperature for six intermediate and low moisture foods.
When to buy a CX2T.
If lab (and AquaLab) temperatures fluctuate by as much as ±5 degrees C daily, water activity readings will vary less than ±0.01 aw. Often, this much uncertainty in sample aw is acceptable, so there is no need to purchase the CX2T. Such variations in ambient temperatures are uncommon. But, if your lab temperature varies to this degree, and you require better than 0.01 aw precision, you may want a CX2T. Most AquaLab customers purchase our standard model CX2.