Comparison Between Water Vapor Sorption Isotherms Obtained Using The New Dynamic Dewpoint Isotherm Method and those Obtained Using The Standard Saturated Salt Slurry Method

Abstract

The Dynamic Dewpoint isotherm method directly determines sample a_w using chilled-mirror technology, while changes in sample weight are tracked gravimetrically. The objective of this research was to compare the Dynamic Dewpoint isotherms to saturated salt slurry isotherms for five materials: dent corn starch, isolated soy protein, microcrystalline cellulose, crystalline sucrose, and corn flakes. The Dynamic Dewpoint isotherms were obtained using the AquaSorp Isotherm Generator between 0.10 and 0.95 a_w at 25°C. Comparison working isotherms were obtained using 12 saturated salt slurries, between 0.064 and 0.973 a_w in desiccators and proximity equilibration cells at 25°C. The Dynamic Dewpoint isotherms exhibited similar sorption behavior to the saturated salt slurry isotherms, except for corn flakes, where there was a marked decrease in moisture content compared to the saturated salt slurry isotherms between 0.40 and 0.70 a_w. This difference was attributed to the slow diffusion of water into the very dense laminated corn flake matrix, highlighting the rate of sorption (time) dependency of the Dynamic Dewpoint isotherm method. The Dynamic Dewpoint isotherm method offers the opportunity for real-time investigation of water sorption-related material properties, such as the glass to rubbery transition, recrystallization, hydrate formation, and deliquescence.

Keywords:

• Isotherm
• Water activity
• Saturated salt slurries
• Dynamic Dewpoint isotherm
• AquaSorp Isotherm Generator

INTRODUCTION

Water is a key component involved in numerous facets of food systems, such as processing, occurrence of phase transitions, chemical, biochemical and physical stability, sensory perception, and, most importantly, microbial safety. Water activity (a_w), which is defined as the ratio between the vapor pressure of water in equilibrium with a sample and the vapor pressure of pure water at the same temperature and atmospheric pressure, is used to express the availability of water in a food system. Moisture content of a material (usually expressed as a mass of water per unit mass of dry material) plotted as a function of a_w (also termed the relative vapor pressure or food relative humidity) at a constant temperature yields a very useful assessment tool identified as the moisture sorption isotherm. A moisture sorption isotherm relates the amount of water sorbed (either adsorption or desorption) at equilibrium (or steady state) by a material at a given a_w (or relative humidity, %RH) at a constant temperature. Sorption isotherms are used for a variety of processing and product stability applications, such as prediction of moisture transfer, development of new products, determination of product stability and shelf life, and process design and control.[1,2]

Traditionally, isotherms have been generated by placing the food material over saturated salt slurries with specific temperature dependent a_w values in closed chambers (e.g., desiccators) and waiting until the food material attains a steady state, that is, reaches the same a_w as the saturated salt slurry, as shown by minimal weight change. The advantage of this method is that the saturated salt slurries generate accurate and precise a_w (or %RH) values as a function of temperature and, if made properly, act as “infinite a_w sinks.” However, Levogeur and Williams[3] have identified four main issues with employing the standard saturated salt slurry method: (1) the lengthy period of time it takes the material to achieve equilibrium; (2) the difficulty of obtaining conclusive, accurate measurements due to the protocol of continuously removing the sample from the chamber and exposing it to the environment, which is often at a different %RH, and possibly temperature, than the sample;[4,5] (3) the requirement of using large sample sizes (typically greater than one gram); and (4) the time-consuming, cumbersome steps required of the researcher to obtain weight measurements during the equilibration period. Additional drawbacks include the finite number of a_w values available (only those associated with specific saturated salt slurries) and the potential mold growth problems during lengthy equilibration times at a_w values ≥0.70.[6]

These drawbacks associated with the saturated salt slurry method have, in large part, been overcome by the recent development of automated dynamic water vapor sorption instruments. A number of these automated instruments operate by continuously monitoring sample weight changes and controlling %RH at select temperatures by adjusting a mixture of dry and water vapor saturated gas streams using mass flow controllers. These instruments are usually programmed to automatically change the %RH in a stepwise fashion, progressively obtaining weight change data used for determining an isotherm on the same sample in a reduced amount of time (compared to the saturated salt slurry method[5]), while virtually eliminating sample handling by the researcher and exposure to room conditions during sample weight measurements. The sample is held at each selected %RH value until the sample weight does not change within a narrow, pre-specified tolerance level before moving to the next %RH value. Some commercially available instruments include the Dynamic Vapor Sorption (DVS) instrument from Surface Measurement Systems Ltd (Middlesex, UK), the IGAsorp from Hiden Analytical (Warrington. UK), the VTI Sorption Analyzer and the Q5000SA from TA Instruments (New Castle, DE, USA), the CI Sorp Water Sorption Analyzer from CI Electronics Limited (North Brunswick, NJ, USA), the SPS Moisture Sorption Analyzers from Projekt Messtechnik (Ulm, Germany), the HydrosorbTM 75 1000 Water Vapor Sorption Analyzer from Quantachrome Instruments (Boynton Beach, FL, USA), and the HygroGen 2—Precision Humidity and Temperature Generator from Rotronic (Hauppauge, NY, USA). As a group, these instruments and isotherms so obtained can be termed controlled atmosphere microbalances (CAM) and CAM isotherms. The interested reader is referred to a study by Rahman and Al-Belushi[7] comparing the saturated salt slurry method (static) to the dynamic CAM method (specifically the Symmetrical Gravimetric Analyzer [Model 100] by VTI, Hialeah, FL, USA).

Another new automated dynamic water vapor sorption instrument was recently introduced, the AquaSorp Isotherm Generator (Fig. 1, Decagon Devices Inc., Pullman, WA, USA), which uses the Dynamic Dewpoint Isotherm (DDI) method for obtaining isotherms.[8] The DDI method differs from the CAM isotherm methods described above, as the %RH of the gas stream in contact with the sample is not controlled. Rather, a_w is directly measured using a chilled mirror dewpoint sensor, while changes in sample weight are tracked gravimetrically using a magnetic force balance. Adsorption (or wetting) is accomplished by saturating the air with water before it enters the sample chamber and desorption (or drying) is achieved by passing the air through desiccant before it enters the sample chamber, both at a user-selected flow rate. ¹ After a small change in a_w, approximately 0.015, airflow is halted and a snapshot of the sorption process is obtained by directly measuring sample a_w and weight, which is subsequently converted to moisture content. DDIs are obtained very rapidly (e.g., several hours to a few days) because the sample is not required to equilibrate to a known %RH value; rather, the sample a_w is directly measured at each successive adsorption or desorption point, allowing for the collection of a large number of a_w points, often >75 data points, over a sizable a_w range (0.03 to 0.95 a_w) and temperature (15 to 40°C) range. Because samples are not required to come to equilibrium in the DDI method, the resultant isotherm may or may not be similar to the saturated salt slurry or CAM isotherms described above. Whether the resultant DDIs are similar to the saturated salt slurry or CAM isotherms depends on the nature of the sample (e.g., state of the material [crystalline or amorphous], particle size, porosity, density, and spatial distribution of the components), the water vapor diffusion rates into or out of the material being investigated (as influenced by the material properties and the selected flow rates), and the presence of any physical changes that may occur in the material, such as the glassy to rubbery transition, recrystallization, hydrate formation, and deliquescence.

Figure 1 Schematic diagram of the AquaSorp Isotherm Generator used to obtain Dynamic Dewpoint isotherms.

To date, a few published research papers have employed the DDI method to obtain sorption isotherms,9–12 however, no publications were found that compare the DDI method to the standard saturated salt slurry isotherm method. Thus, the objective of this research was to obtain DDIs of five diverse food materials, dent corn starch, isolated soy protein, microcrystalline cellulose, crystalline sucrose, and corn flakes, and compare them to previously obtained saturated salt slurry isotherms using a desiccator (DES) and proximity equilibration cell (PEC) chambers.

MATERIALS AND METHODS

Materials

Five samples, normal dent corn starch (Tate & Lyle, Decatur, IL, USA), isolated soy protein (ADM, Decatur, IL, USA), microcrystalline cellulose (FMC Corporation, Philadelphia, PA, USA), crystalline sucrose (Sigma-Aldrich, St. Louis, MO, USA), and corn flakes (medium size whole flakes [average size 1.8 by 1.1 cm]; Kellogg's, Battle Creek, MI, USA), were used in this study “as is.” The crystalline sucrose particle size was measured using USA standard testing sieves (W.S. Tyler, Mentor, OH, USA), where 98.2% of the sucrose was greater than 425 μm and all particles were greater than 300 μm. Samples were stored in their original containers, and the containers were placed into sealable low water vapor transmission rate plastic bags at room temperature and relative humidity conditions.

The “as is” dent corn starch, isolated soy protein, microcrystalline cellulose, and corn flakes moisture contents were determined by placing triplicate samples in a vacuum oven (30 in. vacuum mercury) at 60°C for 24 h. The “as is” crystalline sucrose moisture content was determined in triplicate using Karl Fisher titration. The “as is” a_w values of all five materials were determined in triplicate at 25°C using the AquaLab series 3 water activity meter (Decagon Devices, Pullman, WA, USA). The average “as is” moisture content values, a_w values at 25°C, and associated standard deviations for each material are given in Table 1. In addition, the average starting DDI a_w values, estimated by the AquaSorp Isotherm Generator (Decagon Devices Inc., Pullman, WA, USA) and associated standard deviations are also listed in Table 1. The DES and PEC methods were carried out by Yu,[5] whereas the DDI method was carried out a few months later on the same lot of the materials (as stored in the lab as described above).

Table 1 Average “as is” moisture content and AquaLab measured a_w values and associated standard deviations at 25°C for the materials used in the DES and PEC and the DDI methods.* The average starting DDI a_w values for both adsorption and desorption parts of the working isotherm, estimated by the AquaSorp Isotherm Generator, and associated standard deviations are also listed. Visually observed mold growth for some of the materials in both the DES and PEC chambers is also indicated. No visual mold growth was observed in the DDI method

Material	MC (% wb)	AquaLab aw values (25°C)	Average starting DDI aw values (25°C)	Mold growth in DES and PEC chambers
Dent corn starch	9.2 ± 0.12	0.447 ± 0.001	0.484 ± 0.026	None
Isolated soy protein	3.8 ± 0.06	0.207 ± 0.001	0.290 ± 0.024	DES ≥ 0.843 aw PEC ≥ 0.903 aw
Microcrystalline cellulose	2.3 ± 0.01	0.408 ± 0.002	0.247 ± 0.039	None
Crystalline sucrose	0.04 ± 0.0003	0.290 ± 0.001	0.313 ± 0.202	PEC ≥ 0.903 aw
Corn flakes	3.0 ± 0.06	0.205 ± 0.004	0.376 ± 0.139	DES and PEC ≥ 0.843 aw
*DES method = saturated salt slurry isotherms obtained using desiccator chambers; PEC method = saturated salt slurry isotherms obtained using proximity equilibration cell (PEC) chambers; DDI method = dynamic dewpoint isotherms obtained using an AquaSorp Humidity Generator.

Methods

Saturated salt slurry method in DES and PEC chambers

Saturated salt slurry working isotherms were obtained in triplicate for 5 materials using both DES and PEC[13] chambers at 12 saturated salt slurry a_w values: 0.064, 0.113, 0.225, 0.328, 0.432, 0.576, 0.689, 0.753, 0.843, 0.903, 0.936, and 0.973 at 25°C.[5] All saturated salt slurry a_w values were from Greenspan,[14] except for 0.903, which was from Young.[15] Samples (1 to 2 g) were weighed at 7-day intervals until each sample did not change by more than 2 mg/g dry weight for two successive weighing times.[16] The average time for a material to reach equilibrium was weeks to months in the DES chambers and days to weeks in the PEC chambers. For example, it took between 2 and 10 weeks to obtain isolated soy protein and corn flake isotherms using the DES chamber and between 1.5 and 2 weeks using the PEC chamber, with specific times dependent upon material type and saturated salt slurry a_w value. After the equilibrium weight had been reached, the percent dry basis (%db) equilibrium moisture content was calculated using the initial and final weight readings and the initial measured moisture content values (Table 1). Mold growth was visually observed for some of the materials in both DES and PEC chambers and is recorded in Table 1. Since weight change data from samples with mold growth are unreliable,[6] these data were excluded from the DES and PEC isotherm plots. No visual mold growth was observed in the DDI method.

DDI Method

The Dynamic Dewpoint isotherms (DDIs) were obtained using the AquaSorp Isotherm Generator (Decagon Devices Inc., Pullman, WA, USA). Before the DDIs were obtained, the a_w dewpoint sensor was verified using two unsaturated aqueous salt solutions purchased from the instrument manufacturer, 13.41 molal LiCl (0.250 ± 0.003 a_w) and 6.0 molal NaCl (0.760 ± 0.003 a_w). Working DDIs were obtained for each material (1 to 2 g in a stainless steel sample cup half filled, except for crystalline sucrose where the sample covered the bottom of the cup; internal radius 1.87 cm, internal height 0.973 cm, and sample volume 10.7 cc) between 0.10 to 0.95 a_w (except for crystalline sucrose that was brought to 0.93 a_w) using the start from current a_w feature (AquaSorp firmware version ASIG 1.19, Decagon Devices, Pullman, WA, USA) with a 300 ml/min flow rate at 25°C. The adsorption and desorption sections of the working isotherm were obtained in triplicate, each on a different sample. The average DDI starting a_w and associated standard deviation for each material for both adsorption and desorption parts of the working isotherm are given in Table 1. Weight change at each a_w was converted to moisture content (%db) using the SorpTrac software (version 1.03.3, Decagon Devices, Pullman, WA, USA) using the initial measured moisture content value (Table 1). The average DDI starting a_w is given since overlap sometimes occurs between the adsorption and desorption isotherm sections of the DDI. This overlap ² may be attributed to samples being run continuously, leaving a “humidity memory” in the sample chamber, which sometimes affects the initial conditions of the subsequent sample.

RESULTS AND DISCUSSION

Triplicate measurements were plotted for all isotherm methods (except for crystalline sucrose, where only a single DDI was plotted, since plotting all three DDI, which fell right on top of each other, made it impossible to observe the higher a_w DES isotherms points), so the variation in the isotherm methods could be visually assessed. Statistical analysis was not carried out between the two isotherm methods for all five materials, since, as discussed previously, the DDI method does not always yield an isotherm that is comparable to traditional equilibrium isotherms; thus, the DDI method should be viewed as a unique isotherm method.[8,17] As a unique isotherm method, some advantages as well as limitations are expected. In regards to its limitations, for example, in the case where the DDI is not comparable to the traditional equilibrium isotherm, isotherm equations, such as the BET and GAB, can still be applied to the DDI data; however, the resultant parameters (e.g., the monolayer moisture content value) are probably not comparable to the isotherm equation parameters associated with the equilibrium isotherm data.

In regards to its advantages, the DDI provides a new opportunity to probe time dependent water sorption-related material properties, such as the glassy to rubbery transition, recrystallization, hydrate formation, and deliquescence. In cases where the DDIs are similar to the saturated salt slurry isotherms, statistical analysis may be done. However, since the saturated salt slurry method yields discrete a_w points and the DDI method near continuous a_w points with each run yielding a unique set of a_w values, formal statistical analysis is not possible; rather the DDI moisture content values may be globally compared to the calculated and, subsequently, connected confidence interval for the discrete DES and PEC isotherm moisture content values. In this study, a global statistical comparison between the two isotherm methods was done for microcrystalline cellulose.

Dent Corn Starch

The dent corn starch DDI (Fig. 2) exhibited similar sorption behavior to its corresponding saturated salt slurry isotherms. This indicates that dent corn starch exhibits relatively fast water vapor diffusion; that is, water vapor penetrates rapidly into or out of the entire sample yielding a DDI that is comparable to the DES and PEC saturated salt slurry isotherms. There, however, appears to be a slight difference between the dent corn starch DDI and the DES and PEC isotherms near 0.576 a_w, indicating that a time dependent change in the sorption process may be occurring in the dent corn starch near this a_w. It is interesting to note that an earlier DDI run, done on the same dent corn starch, did not exhibit this mid-isotherm dip.[8] The appearance of the mid-isotherm dip in some DDI runs and not others, further suggests the occurrence of some type of time dependent physical change in the material that is perhaps related to the initial starting conditions of the material (e.g., moisture content and a_w) and/or the age and history of the sample. The exact nature of this time dependent physical change is difficult to identify, since native (raw) dent starch is a complex biomaterial, composed of granular structures containing both amylose and amylopectin polymers, with both crystalline and amorphous regions. Pregelatinized starch (no crystalline regions) may be a better material choice to use for comparing these isotherm methods and for further investigation into the nature of the slight difference in the isotherms obtained by these two methods.

Figure 2 DES, PEC, and DDI isotherms for dent corn starch (color figure available online).

Isolated Soy Protein

The isolated soy protein DDI (Fig. 3) was similar to the corresponding saturated salt slurry isotherms, however, the DDI exhibited slightly lower moisture contents near 0.432 a_w. This difference in the isotherms may be attributable to the occurrence of a glassy to rubbery transition in the amorphous portion of the isolated soy protein. However, additional research is required to confirm this hypothesis. The Savitsky Golay second derivative method[18] was applied to the DDI a_w data,[17] yielding 0.458 as the critical a_w (or critical %RH) at the experimental temperature, 25°C. If our hypothesis is correct, this critical a_w corresponds to the glass transition temperature of the material for the experimental conditions employed. However, it is important to keep in mind that the critical a_w at which the glassy to rubbery transition occurs at a given temperature (the temperature of the experiment that corresponds to the glass transition temperature) will most likely be dependent on the time scale under which the DDI measurements are obtained (e.g., flow rate and a_w measurement time) as well as the nature of the sample. The difference in the measurement time between the DES and PEC (long time) and the DDI (short time) isotherms may allow changes in the physical state of the material to occur to different extents, which are then reflected in the isotherms produced by the two different methods. The low data resolution in the standard saturated salt slurry methods makes it very difficult to observe these sorption driven transitions. Further investigation into the relationship between the critical a_w as determined via the DDIs and the glassy to rubbery transition is currently being carried out using polydextrose as a model system.[19]

Figure 3 DES, PEC, and DDI isotherms for isolates soy protein. Mold growth a_w limits for DEC and PEC methods are given in Table 1 (color figure available online).

Microcrystalline Cellulose

The microcrystalline cellulose DDI (Fig. 4) exhibited similar sorption behavior to its corresponding saturated salt slurry isotherms, indicating that microcrystalline cellulose exhibits relatively fast water vapor diffusion. Since the DDI appears similar over the entire a_w range to the saturated salt slurry DES and PEC isotherms, a global statistical comparison of one ³ of the working DDIs to the connected upper and lower 99% DES/PEC confidence limits (±0. 506 mc %db) is shown in Fig. 5. As can be seen in Fig. 5, the DDI data remained in between or on (although, in general, nearer to the lower confidence limit) the 99% DES/PEC confidence limits, except for the last two high a_w DDI points, indicating that overall, the DDI, in the case of MCC, was the same as the DES and PEC isotherms. It is not possible to carry out a formal statistical analysis since the saturated salt slurry method yields discrete a_w points and the DDI method near continuous points with each run yielding a unique set of isotherm points.

Figure 4 DES, PEC, and DDI isotherms for microcrystalline cellulose (color figure available online).

Figure 5 Informal statistical comparison of one of the working DDIs to the connected upper and lower 99% DES and PEC confidence limits (±0. 506 mc %db) (color figure available online).

It is interesting to note that for microcrystalline cellulose the two sections of the working isotherm that overlapped (the desorption and the adsorption sections) were further apart than for the other materials; that is, the moisture contents were slightly different at the same a_w (around the 0.20 to 0.30 a_w range). Further investigation is needed to determine the cause of this gap. However, it is also interesting to note that the moisture contents for the DES and PEC isotherms in the 0.20 to 0.30 a_w range (0.225 a_w) also exhibited a larger variation, compared to the moisture contents at other nearby a_w values.

Crystalline Sucrose

The crystalline sucrose DDI (Fig. 6) was also similar to the corresponding saturated salt slurry isotherms; however, the DDI was able to completely capture the deliquescence process, whereas the saturated salt slurry method only allows for isotherm values to be obtained at a few discrete a_w values. The deliquescence point, as determined by taking the average of the a_w values before and after the turning point, was 0.852 ± 0.001 at 25°C, which is virtually the same as the value reported by Salameh and others[20] of 0.85 at 25°C obtained by extrapolation using a VTI Sorption Analyzer (VTI, Hialeah, FL, USA). Comparing the deviation of the upper vertical DDI to the saturated sucrose solution a_w and moisture content values plotted in Fig. 6 shows that during the extensive water sorption that accompanies the deliquescence process, a slightly less than saturated sucrose solution layer begins to form, which is subsequently measured by the dewpoint sensor. This is due to the dynamic, diffusion-limited nature of the DDI measurement method. Despite the formation of this less than saturated sucrose solution layer, the DDI method provides one of the most insightful tools for observing the deliquescence process, as well as one of the most accurate means of determining the deliquescence point. Further discussion of the use of the DDI method to determine the deliquescence a_w (or %RH) is given in Yao and others.[12]

Figure 6 DES, PEC, and DDI isotherms for crystalline sucrose. The deliquesce point and the literature saturated sucrose solution a_w (at 25°C from Salameh et al.[20]) and associated moisture content (at 25°C from Bubnick and Zadlec[21]) are indicated on the figure. Mold growth a_w limits for DEC and PEC methods are given in Table 1 (color figure available online).

Corn Flakes

The corn flake DDI (Fig. 7) was substantially different than the corresponding saturated salt slurry isotherms in the a_w region between 0.40 and 0.70. The Savitsky Golay[18] second derivative yielded 0.673 as the critical a_w at the experimental temperature, 25°C. The difference in the isotherms is attributed to slow diffusion of water into (absorption) the very dense laminated corn flake matrix, which shows up as the difference between the rapidly obtained DDI and the more slowly obtained equilibrium DES and PEC isotherms. The location of the gap between the DES and PEC and DDI isotherms is most likely associated with an a_w-dependent permeability of the very dense laminated corn flake matrix, where the large width of the gap is attributed to the time-dependent nature of the DDI sorption process. At the critical a_w, the permeability ceases to be a limiting water sorption factor and the DDI isotherm again agrees with the DES and PEC isotherms since additional time is no longer needed for the water to diffuse through the laminate surface of the corn flake. Increasing the surface permeability or decreasing the path length of diffusion should decrease the gap between the isotherms. Shands and Labuza[22] recently illustrated these effects by comparing the DDI of whole, coarsely ground, and finely ground corn flakes to the static saturated salt slurry isotherm method. They reported that grinding the corn flakes increased the surface area available for water sorption and decreased the diffusion distance, resulting in the ground corn flake DDIs being similar to the static saturated salt slurry desiccator isotherm method.

Figure 7 DES, PEC, and DDI isotherms for corn flakes. Mold growth a_w limits for DEC and PEC methods are given in Table 1 (color figure available online).

Below 0.40 a_w, it is hypothesized that water is mainly being adsorbed onto (adsorption section of the working isotherm) or desorbed from (desorption section of the working isotherm) the surface of the corn flakes, following Fickian diffusion (as the corn flake exhibits a type III or “J” shaped isotherm) in each of the working isotherm methods, resulting in very little difference in moisture content.[23] As a_w increases above 0.40, permeability of the laminate surface becomes sufficient for increased adsorption of moisture if given sufficient time (the time required being dependent on the permeability and, hence, the a_w). Consequently, over the a_w range of 0.40 to 0.70, DES and PEC isotherms adsorb more water than the DDI, since the DES and PEC methods provide sufficient time to allow moisture movement beyond the surface and into the corn flake matrix. The adsorption of moisture for the DDI method will continue to lag behind until the permeability is high enough (at 0.673 a_w under the conditions of the current experiment) to allow rapid diffusion of moisture into the matrix. At this point, diffusion of water into the corn flake matrix is fast enough that the DDI and DES and PEC methods are once again in agreement (mold growth at a_w values ≥0.843 in the DES and PEC methods precludes additional higher a_w comparisons). It is reasonable to assume under this scenario that anything that increases diffusion into the sample, such as increasing permeability or decreasing particle size (results discussed above^[22]) or decreased flow rate, will increase the amount of water sorbed by the sample in the DDI method and result in better agreement between the sorption methods. These results also suggest the possibility of further investigating the a_w dependency of permeability by utilizing a combination of equilibrium and DDI methods.

CONCLUSIONS

In general, the DDIs exhibited similar sorption behavior to the saturated salt slurry isotherms for the dent corn starch, isolated soy protein, microcrystalline cellulose, and crystalline sucrose, where relatively minor, but noteworthy, deviations were discussed. The DDIs differed more dramatically from the saturated salt slurry isotherms for corn flakes in the a_w transition region, between 0.40 and 0.70 a_w. This difference was attributed to slow diffusion of water vapor into the very dense laminated corn flake matrix. The benefits of the DDI method compared to the saturated salt slurry isotherm method are: (1) increased analysis speed, a few days compared to weeks or months; (2) a dramatic increase in data point resolution, with often >75 data points per isotherm; (3) prevention of mold growth problems experienced for some materials at high a_w values (mm); and (4) the opportunity to explore sample specific, time dependent water sorption-related material properties, such as the glassy to rubbery transition, recrystallization, hydrate formation, and deliquescence. Along with its benefits, the non-equilibrium nature of the DDI method also has its limitations, including: (1) the resultant DDI not always being equal to the equilibrium saturated salt slurry isotherm (as observed for the corn flakes herein); and (2) in the case where limitation one applies the resultant DDI isotherm equation parameters (e.g., obtained by applying the BET or GAB equation to the DDI data) are not equivalent to those equation parameters obtained for the equilibrium isotherm. This last limitation is important since moisture sorption isotherms are frequently used to obtain the monolayer moisture content value. Based on the initial DDI results reported herein, additional, more in-depth experiments are currently being carried out on these and other food materials. Specially, the DDI method is being investigated for its promise as a new method for determining the relative humidity at which the glassy to rubbery transition takes place at a fixed temperature.

ACKNOWLEDGMENTS

This project was supported in part by the USDA National Institute of Food and Agriculture, Hatch project number 385. In addition, the authors gratefully acknowledge the financial support of the 2008 Marcel Loncin Research Prize from the Institute of Food Technologists and the 2008 Midwest Advanced Food Manufacturing Alliance (University of Nebraska-Lincoln) grant. The helpful scientific discussions with Brady Carter of Decagon Devices (Pullman, WA, USA) and Dr. Anthony J. Fontana, Jr. of Silliker, Inc. (Homewood, IL, USA) are also gratefully acknowledged.

Notes

¹The user selected flow rate can range from 10 to 1000 ml/min, where a lower flow rate allows more time for moisture penetration.

²Subsequent to the data collected herein, to minimize this overlap the manufacturer now suggests conditioning the sample chamber between sample runs by running an activated charcoal sample. In addition, the manufacturer added a 30 min stabilization step before the sorption run begins.

³For clarity purposes, only one microcrystalline cellulose DDI was plotted. However, the other two microcrystalline cellulose DDIs behaved similarly.