RAPID WATER Acnvrrv ‘EVALUATEGN OF- DEHYDRATED 2 may mowers ‘~ ‘ Thesis for the aegree of M. s. ' mum STATE uawzasm Vim cums 3000 1932 ‘VHfls ABSTRACT RAPID WATER ACTIVITY EVALUATION OF DEHYDRATED DAIRY PRODUCTS BY Vijay Chand Sood To use water activity as an index of the shelf stability of foods, there is a need for a rapid and accu- rate method to evaluate the water activity of foods. A simple vacuum manometer was designed to measure the water vapor pressure of foods up to 30 C over the complete water activity range. The procedure to evaluate the water activity of dehydrated dairy products was standardized. One minute evacuation and one minute freezing was enough to get reproducible results. Beyond this, the water activ- ity was not affected by the evacuation time or the freezing time. Water activity was independent of the sample size; sample tube volume; and the particle size of the sample. Water activity increased with temperature, and decreased with the repeated use of a sample through freezing— equilibration cycles. Water activity decreased if the reference side of the manometer was also sealed off from Vijay Chand Sood the vacuum pump along with the sample side. The difference was consistent, so the new approach can be used with a reference curve to correct the results so obtained. The method was accurate to within $0.01 water activity with samples equilibrated over saturated salt solutions. The method gave highly reproducible results with a standard deviation of 0.00365 for a market sample and 0.0018 for an equilibrated sample over a saturated salt solution. The time required to reach equilibrium in the apparatus showed a slight increase with a decrease in particle or agglomerate size of the sample. There was a time lag before the vapor pressure started increasing. The time lag increased with a decrease in the agglomerate size fraction of instant nonfat dry milk. The time required to reach equilibrium decreased when a water bath was used in place of equilibration in air. The average time to run a test was 17 minutes for the instant and 30 minutes for the regular spray dried nonfat dry milk. RAPID WATER ACTIVITY EVALUATION OF DEHYDRATED DAIRY PRODUCTS BY Vijay Chand Sood A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Food Science and Human Nutrition 1972 ACKNOWLE DGMENTS The author wishes to express his appreciation to Dr. D. R. Heldman, his major professor, for his encourage- ment and able guidance to complete this work. Thanks are due to Dr. C. M. Stine for the under- standing and help extended during the early part of author's stay in the department. Special thanks are due to Dr. F. W. Bakker—Arkema and Dr. P. Markakis for reviewingthe manuscript and giving helpful suggestions. The author also wishes to express his appreciation to Indian Institute of Technology, Kharagpur, India and the Ford Foundation for giving this Opportunity for academic improvement. ii TABLE OF CONTENTS LIST OF TABLES . . . . . . . . LIST OF FIGURES O O O O O O 0 INTRODUCTION 0 O O O O O O 0 REVIEW OF LITERATURE . . . . . . Sorption Phenomenon in Foods . . Water Activity and Food Stability Measurement of Water Activity . . EXPERIMENTAL . . . . . . . . Design of Apparatus . . . . . Sample Treatments . . . . . . Procedure . . . . . . . . RESULTS AND DISCUSSION . . . . . Standardization of the Method . . Reproducibility of the Method . . Accuracy of the Method . . . . Rate of Equilibration . . . . Recommended Design and Procedure . SUMMARY AND CONCLUSIONS . . . . . REFERENCES . . . . . . . . . iii Page iv vi 16 16 19 21 25 25 37 39 43 51 53 56 LIST OF TABLES Table Page 1. Effect of freezing time on water activity, equilibration time, and pressure due to desorbed gases on nonfat dry milk samples . . . . . . . . . . . . . 27 2. Effect of the temperature of equilibration on water activity, equilibration time, and pressure due to desorbed gases on nonfat dry milk samples . . . . . . . . . . 28 3. Effect of sample size on water activity, equilibration time and pressure due to desorbed gases on nonfat dry milk samples . . . . . . . . . . . . . 30 4. Effect of the volume of the sample tube on water activity of nonfat dry milk samples . . . . . . . . . . . . . 32 5. Effect of the repeated use of a sample through freezing-equilibration cycles on the water activity of nonfat dry milk samples . . . . . . . . . . . 33 6. Effect of the repeated use of a sample through freezing-equilibration cycles on pressure due to desorbed gases of nonfat dry milk samples . . . . . . . 34 7. Effect of sealing both sides of the mano- meter on the water activity of nonfat dry milk samples . . . . . . . . . . 36 8. Water activity data of spray dried nonfat dry milk for statistical analysis . . . . 38 9. Water activity of nonfat dry milk samples equilibrated over Phosphorus pentoxide . . . . . . . . . . . . 39 iv Table Page 10. Water activity of nonfat dry milk samples equilibrated over saturated salt solu- tions under vacuum and at room temperature . . . . . . . . . . . 40 11. Water activity of nonfat dry milk samples equilibrated over saturated salt solu- tions at atmospheric pressure and at room temperature . . . . . . . . . 41 12. Water activity of nonfat dry milk samples equilibrated over saturated salt solu- tions at atmosPheric pressure and at 30 C . . . . . . . . . . . . . 42 13. Water activity data of Spray dried nonfat dry milk sample equilibrated over a saturated salt solution of Potassium acetate at atmospheric pressure and at 30:0.5 C for statistical analysis . . 44 14. Effect of particle size fraction of regular spray dried nonfat dry milk on the time required to reach equilibrium . . . . . . . . . . . 45 15. Effect of agglomerate size fraction of instant nonfat dry milk on water activity and time required to reach equilibrium . . . . . . . . . . . 46 16. L value as a function of the agglomerate size fraction of instant nonfat dry milk 0 O O O O O O O O O O O 49 LIST OF FIGURES Figure l. Vapor pressure apparatus for water activity measurement . . . . . 2. Rate of equilibration as a function of agglomerate size fraction of nonfat dry milk--instant . . . 3. L value as a function of the average agglomerate diameter of nonfat dry milk--instant . . . . . . vi Page 18 48 50 INTRODUCTION Water is a major component of most natural and processed foods and, to a large degree, controls the physicochemical properties and the shelf stability of foods. Water is required as a reactant and as a medium for the transport of the reactants to the reaction sites. If not adequately processed and stored, high moisture foods undergo rapid microbiological and enzymatic spoilage. Re- duction of total water content by Concentration or dehy- dration, is used as one of the methods to preserve foods from biological spoilage. In these foods chemical reac- tions still go on, resulting in slow deterioration of the edible and the nutritional quality of these foods. Shelf stability of foods was generally believed to be a function of total water content. Efforts were, there- fore, concentrated to manufacture dehydrated foods with minimum possible water content. Some foods were, however, found to exhibit an optimum, not a minimum, water content for maximum stability. Also, the minimum water content for different foods was found to be different, for the same degree of food stability. This led to investigations into the mechanism of water binding in foods. Water in foods is bound by the food components to a varying degree. The degree of water binding is reflected in a reduction of the vapor pressure of water in foods. The ratio of the water vapor pressure of a food to the vapor pressure of pure water at the same temperature, is defined as the water activity (aw) of the food. Water activity, therefore, is an index of the relative freedom of water in foods to support biological and chemical reac- tions. Water activity, not the total water content, con- trols the chemical; enzymatic; and microbiological reac- tions in foods at the microenvironmental level. Deve10pment of a rapid and accurate method to evaluate the water activity of foods is needed to use water activity as an index of the shelf stability of foods. The plots of water content versus the water activity, called sorption isotherms, can be used as reference curves to find out the water activity of foods. Sorption isotherms, however, have the following limitations: 1. Deve10pment of sorption isotherms is a laborious and time consuming process. 2. Sorption isotherms are a function of product composition, processing history, temperature and sorption hysteresis. A set of isotherms is, there— fore, required to describe each of these conditions. 3. Complete information about the above mentioned variables should be known to select the correct isotherm as the reference curve for a food. 4. The accuracy of the method cannot be better than the accuracy of the method selected for the moisture determination. Rapid and accurate methods for the moisture determination in foods are not yet available. 5. The method cannot be used for the development of new foods based on the control of water activity. Direct measurement of water activity is a more desirable approach. The relative humidity of the vapor Space in moisture equilibrium with a food sample, ex- pressed as a fraction, is equal to the water activity of the food. The measurement of equilibrium relative humid— ity has not been in much use because sensitive and accu— rate hygrometers were not available. The modern electric hygrometers are more sensitive with a claimed accuracy of :1 percent. These hygrometers are expensive and difficult to use. Measurement of the water vapor pressure of foods in a suitably designed vacuum manometer has been used more extensively. The apparatus is simple, easy to use, re- quires a small sample and is adaptable to routine quality control work. Only limited work has been reported on the accuracy and reproducibility of the method. The objectives of this research were: To design a simple vacuum manometer to measure the water vapor pressure of foods. To standardize the method to measure the water vapor pressure of foods. To investigate the influence of the following experimental factors: a. Time for evacuation. b. Time for freezing. c. Temperature of equilibration. d. Air versus water equilibration. e. Sample size. f. Sample tube capacity. 9. Repeated use of a sample. h. Sealing both sides of the manometer. To study the effect of particle size on rate of equilibration. To test the reproducibility of the method. To test the accuracy of the method. REVIEW OF L I TERATURE Sorption Phenomenon in Foods Sorption phenomenon in foods refers to the exchange of moisture between foods and the vapor space surrounding the foods. The amount of moisture held by a food is a function of the water binding properties of the food com- ponents, temperature and the relative humidity of the vapor Space. Sorption continues until equilibrium is attained between the water vapor pressure of the food and the water vapor pressure of the vapor Space. The ratio of the equilibrium vapor pressure to the vapor pressure of pure water at the same temperature, is called the water activity of the food. Water activity, therefore, is an index of the degree of water binding by foods. The plots of water content aS a function of water activity are called sorption isotherms. Rockland (1969) concluded that the moisture sorption isotherms of hetero- geneous biological products represent the integrated hygroscopic prOpertieS of numerous constituents which vary in respect to both quality and quantity. Salwin (1963) found that starchy foods have the greatest water holding capacity, followed by foods rich in proteins. Foods with high sugar content showed much lower water holding capac- ity. Taylor (1961) reported that fats in meats exerted negligible water vapor pressure. Presence of fat, how- ever, did Slow equilibration, tending to coat the parti- cles of the sample and prevent free transfer of moisture. Labuza (1968) mentioned that pretreatment like heating has little effect on proteins, but the water holding capacity of starches is reduced by changeover from amorphous to the crystalline state. Salwin (1959) found that exchange of moisture between dehydrated foods tended towards equal water activity, not equal water content. Labuza (1968) reviewed the sorption phenomenon in foods and attributed the sigmoid character of the sorption isotherms of foods to three types of bound water in foods. The sorption isotherms could be divided into three regions correSponding to adsorption of a monomolecular film of water; adsorption of additional layers on the monolayer; and condensation of water in the pores of the food mater— ial, respectively. The author also concluded that the three regions overlap each other, limiting the usefulness of any one mathematical model for describing the whole isotherm. Rockland (1969) presented further evidence of three types of bound water from nuclear magnetic resonance; electron spin resonance; and phOSphorescence decay studies on gelatin equilibrated to different moisture contents. The author concluded that the three types of bound water has been associated with molecular groups rather than chemical Species: Type 1 water binding is regarded as water mole- cules bound to ionic groups, such as carboxyl and amino groups. Type 2 water binding is assumed to consist of water molecules hydrogen bonded to hydroxyl and amide groups. Type 3 is considered as unbound, free water found in interstital pores in which capillary forces and solu- ble constituents cause vapor pressure lowering consistent with Raoult's law. Sorption isotherms for foods can be developed either by adsorption or desorption process. Adsorption and desorption isotherms show a hystersis loop. Stitt (1958) stated that the magnitude of hysteresis is deter- mined primarily by food composition, but also depends on the temperature and pretreatment of the sample. Labuza (1968) reviewed the theories on hysteresis and found the 'ink bottle theory' most acceptable. The theory assumes capillaries composed of narrow necks and large bodies. The larger radius controls adsorption and the smaller radius controls desorption, producing the hysteresis loop. Rockland (1969) believed molecular Shrinkage during desorption reduces sorptive sites on the adsorbent surfaces, producing a metastable state, resulting in hysteresis. Water Activity and Food Stability Food stability is not an absolute concept, but refers to the relative resistance of foods to chemical and biological deterioration during handling and storage. The general belief that food stability is a function of total water content, is no longer valid. Water activity, which is an index of the degree of water binding by foods, has been Shown to be more closely related to food stability than the total water content. Salwin (1959) examined the moisture levels required for stability of dehydrated foods. The author found that moisture content corresponding to adsorption of a mono- molecular film of water appeared to be the most stable water content. Lipid oxidation increased below the monolayer value and nonenzymatic browning increased above the mono- layer value. Labuza 35 El. (1969) attributed the protective ef- fect of water on lipid oxidation to two factors. One, hydration of metal catalysts decreasing their effectiveness. Two, hydrogen bonding of peroxides stOpping the free radical chain reaction. The protective effect was evident up to water activity of about 0.5, beyond which the rate of oxi- dation increased in the capillary region. Labuza gE_al. (l972,a) attributed the increase to increased catalyst mobility and swelling of the polymeric matrix, exposing new catalyst sites. Karel and Labuza (1968) found the rate of nonenzy- matic browning increased with an increase in water activity up to the intermediate moisture range and then decreased. The increase was attributed to the increased solubilisation and mobility of the reactants. The decrease was attributed to dilution of the reactants, after all the reactants had been solubilised, according txa law of mass action. Acker (1969) found a Significant dependence of enzyme reactions on water activity, not water content. This View was supported by the fact that oxidases, which do not use water as a reactant, showed the same dependence as hydrolases. Lack of free water below the monolayer value prevented the diffusion of substrate to the enzyme, result- ing in no or very low reaction rates. This was not true for nonaquous liquid substrates which could freely move to the enzyme and reactions could occur at much lower water activity. Scott (1957) reviewed the water relationships of food spoilage microorganisms. The author concluded that the reduction of water activity below an optimum level caused an increase in the lag period; a decrease in the rate of growth; and a decrease in the amount of cell struc- ture synthesized. These results were very similar to those produced by reducing the temperature below the optimum 10 level. It was predicted that each organism was likely to have its own characteristic Optimum water activity for most rapid growth. The author summarized that the fundamental importance of water activity had been strongly supported by experiments Showing that the biological reSponse to a certain water activity was, at least for some organisms, largely independent of the type of solute and the total water content of the substrate. Bone (1969) tabulated the approximate lower limits of water activity, at which growth could occur for various types of microorganisms. This knowledge has been extensively used in developing the technology of intermediate moisture foods. Rockland (1969) stated that optimum stability condi- tions are described more precisely by the combination of equilibrium relative humidity and total moisture content coordinates of a specified moisture sorption isotherm. The author suggested that the stability of processed or blended food products may be improved by appropriate formula modi- fications consistent with the expansion of the local iso- therm II range, corresponding to the type 2 bound water in foods. Labuza gt El. (1972, a and b) examined the chemical and microbiological stability of intermediate moisture foods. It was found that foods prepared by an adsorption process Showed much better stability than the foods 11 prepared by a desorption process. The authors concluded that water activity was not the only factor controlling food stability, but one must think in terms of water activ- ity and total water content. Foods prepared by the desorp- tion process were found to contain more moisture compared to foods prepared by the adsorption process, for the same water activity. Labuza EE.E£° (l972,a) presented a stability map of foods Showing reaction rates as a function of water activ- ity. This stability map can be used as a general reference to predict the type and rate of deterioration that a food of known water activity is likely to undergo during handling and storage. Measurement of Water Activity Sorption isotherms for foods are generally developed by exposing foods to controlled humidity environments and measuring the water content when equilibrium is attained. An alternative approach is to measure the water activity of foods at different water contents. Direct measurement of water activity of foods is based on two principles. 1. The relative humidity of the vapor space in mois- ture equilibrium with a food, expressed as a frac- tion, is equal to the water activity of the food, by virtue of the equilibrium condition. 12 2. By definition, the water vapor pressure of a food divided by the vapor pressure of pure water at the same temperature, is the water activity of the food. Mossel 3E 31. (1955) found that hair hygrometers had a very low sensitivity and reliability. These re- searchers adopted a lithium chloride cell to measure the relative humidity of the vapor spaces above saturated salt solutions and some food products. The device was reported to give highly reproducible results. Rockland (1957) devised an A.C. type electric hygrometer equipped with a multielement humidity and temperature sensing unit. The unit had a magnetically driven fan, located below the base of the sensing element, for faster equilibration of the samples. The author re- ported good agreement with saturated salt solutions. Labuza EE.E£° (l972,a) reported very variable results with an electric hygrometer. More success has been achieved with direct water vapor pressure measurement of foods. The differential vacuum manometer after Gibson and Adams (1933) is a stand- ard apparatus used by physical chemists to measure the vapor pressure of solutions. The apparatus is evacuated to a low pressure of about one micron and the vapor pres- sure of the solution is measured on a sensitive oil mano- meter. The reference pressure of one micron in one arm of the manometer is assumed to be negligible, without 13 introducing a significant error. Some loss of moisture is inevitable during evacuation of the apparatus. The concentration of the solution is, therefore, determined after the vapor pressure measurement. Makower and Myers (1943) designed a Simple appara- tus to measure the vapor pressure of foods. A freeze trap was introduced on the sample side to prevent the escape of moisture during evacuation of the apparatus. At the end of the evacuation period, the manometer was brought into operation and the freeze trap was removed. The moisture in the trap evaporated quickly and readsorbed on the food. Equilibrium vapor pressure was attained in 0.5 to 2.0 hours. The effect of desorbed gaSes during the equilibra- tion process was eliminated by replacing the freeze trap, and measuring the residual pressure in the manometer. The residual pressure was subtracted from the original vapor pressure reading. The authors recommended the vapor pres— sure measurement as an alternative to moisture content determination, not as an index of the Shelf stability of foods. The vapor pressure measurement was reported to be easy, required less time, and gave highly reproducible results. The results were independent of sample size and particle Size, unlike moisture determinations. Taylor (1961) modified the apparatus mentioned above to develOp desorption isotherms for a variety of dehydrated foods. The freeze trap was eliminated, instead M TEE? 14 the sample was frozen to prevent escape of moisture during evacuation of the apparatus. The apparatus was expanded to allow vacuum distillation of moisture from the sample in steps, so that same sample could be used for vapor pressure measurement at different moisture contents. The author recommended grinding of samples for faster equili- E bration, and to keep the volume of the sample side to a minimum to discount any possibility of a reduction in vapor pressure of the sample due to evaporation of mois- ture to fill the vacuum space. The method required about one hour to reach equilibrium. Karel and Nickerson (1964) found the vapor pressure measurement more reliable than equilibration over satur- ated salt solutions at water activities below 0.2, for developing sorption isotherms for dehydrated orange juice. Labuza 22 El. (l972,a) obtained more reproducible results with the vapor pressure technique compared to an electric hygrometer. Elvanides and Markakis (1971) designed an appara- tus to eliminate the use of a vacuum pump for evacuation of the apparatus. A specially designed three-way stOpcock was used to evacuate the sample holder by repeated suction into a barometric vacuum and discharge to atmosphere till no further drOp was observed in the barometric vacuum. The sensitivity of the method was reduced because a mercury manometer had to be used. The process of 15 evacuation is laborious and time consuming. The method, therefore, is not adaptable for a routine quality control use. Good agreement was reported for a variety of foods equilibrated over saturated salt solutions. Slightly higher values at lower water activities, and slightly lower values at higher water activities, were thought to be due to incomplete equilibration of the samples. A Single determination required about 2.5 hours. EXPERIMENTAL Design of Apparatus Water activity varies with temperature, and should be evaluated at the most likely temperature of storage to correctly predict the Shelf stability of foods. Processed and dehydrated foods are generally stored at temperatures around 30 C or lower. The apparatus was designed to measure the water vapor pressure of foods up to a tempera- ture of 30 C, over the complete water activity range. The water activity of pure water is one, and the vapor pressure of pure water at 30 C is 31.824 mm of mercury. Apeizon B oil was used as the manometric fluid which gave a magnification factor of 15.7 compared to mer- cury (Taylor, 1961). The height of the manometer required was 31.824 x 15.7 = 499.637 mm. The manometer could be read accurately up to 0.2 mm, giving a sensitivity of 0.0004 water activity at 30 C. The sensitivity of the apparatus could be increased by using higher temperatures, provided the water vapor pressure of the food did not ex- ceed the vapor pressure of pure water at room temperature. 16 17 This results in local condensation of moisture in the mano- meter (Taylor, 1961). The apparatus is shown in Figure 1. It is a simpli- fied version of the design used by Taylor (1961). It had a 500 mm high oil manometer (A), connected to a three-way ground glass stopcock (B). One arm of the manometer was connected to another three-way ground glass stopcock (C). The bulb of the stopcock (C) was blown and joined to the inner part of a 10/30 ground glass joint to accept the sample tube (D). The manometer was fabricated from 5 mm outer diameter, thick wall glass tubing. The manometer was filled with oil through the stopcock (B) up to half the height. The stopcock (B) was connected to a two-stage vacuum pump of 21 litres per minute free air displacement and 0.1u ultimate pressure attainable. The three sample tubes used to vary the volume of the sample side of the manometer had the following dimensions: Volume Diameter Length 1. 5 ml 13 mm 60 mm 2. 10 ml 24 mm 45 mm 3. 15 ml 24 mm 65 mm A mixture of acetone and dry ice in a Dewar flask was used as the freeze trap. The freezing mixture had a temperature of -80 C. Water has a vapor pressure of 0.4u at -80 C, which is much lower than the pressure to which the apparatus was evacuated. This eliminated the 18 A. Manometer. B. Three way stopcock. C. Three way stopcock. D. Sample tube. __...__, -___._. ,wflmww. .__ __.____. ___:l (C- b ¥ .. Scale 1:5 Figure l.--Vapor pressure apparatus for water activity measurement. l9 possibility of moisture loss during evacuation, provided the sample was frozen for sufficient time to lower the temperature to this level. The samples were originally equilibrated to the room temperature in the air. Later, a constant temperature water bath prepared in a Dewar flask was used. The temp- erature was manually controlled to within $0.1 C by mixing small amounts of water periodically as required during equilibration. Sample Treatments Regular spray dried nonfat dry milk was used as a model system for these studies. _The sample was obtained from the Michigan State University dairy plant. The mois- ture content of the sample was 3.9 percent. The sample was stored in moisture proof glass bottles to check the adsorption of moisture from the atmosphere, as far as pos- sible. It was impossible to completely eliminate the adsorption of moisture during the repeated sampling re- quired for analysis. This has been kept in mind in inter- preting the results. The sample was used without any treatment to stand- ardize the method and to test its reproducibility. To test the accuracy of the method and to study the effect of particle Size on the rate of equilibration, the sample was sieved through U.S. standard sieves on a Ro-Tap Sieve 20 shaker for 5 minutes. The following three major size fractions were collected: 1. 44 to 63 u 2. 63 to 88 u 3. 88 to 125 u A 10 gram sample from each size fraction was uni- formly spread in 85 mm diameter plastic petri dishes. The samples were equilibrated to a constant weight in constant humidity environments in vacuum desiccators. The water vapor pressure of the samples was measured after equili- bration was complete. The following saturated salt solu- tions were used to maintain the constant humidity environ- ments (Wink and Sears, 1950): l. Lithium chloride (LiCl)--ll.2% relative humidity at 30 C. 2. Potassium acetate (CH3COOK)——22.0% relative humidity at 30 C. 3. Magnesium chloride (MgC12)-—32.4% relative humid- ity at 30 C. A set of samples was also equilibrated over Phos- phorus pentoxide, to test the apparatus at zero moisture content in the samples. A commercial sample of instant nonfat dry milk was used to study the effect of large particle agglomerates on rate of equilibration. The sample was sieved through U.S. 21 standard sieves on a Ro-Tap sieve shaker for 5 minutes and the following size fractions were collected: 1. < 125 u 2. 125 to 177 u 3. 177 to 250 u 4. 250 to 500 u 5. 500 to 707 u 6. 707 to 1000 U 7. 1000 to 1410 u The vapor pressure was recorded every two minutes, and the data so collected was analyzed for rate of equili- bration. Procedure Standardization of procedure was one of the objec- tives of this research. The recommended procedure is given at the end of the section on results and discussion. The basic steps of the procedure in general were as follows: 1. The ground glass stOpcocks (B) and (C) were thor- oughly cleaned and dried. A high vacuum grease was applied to the stOpcocks, and the apparatus was tested for any possible leaks. 2. The sample to be tested was quickly transferred to the sample tube (D) to avoid exchange of moisture with the atmOSphere. A funnel was used to transfer the sample to the sample tube to protect the neck 22 of the sample tube from adhering sample particles which could result in a leaky seal. The sample tube was connected to the apparatus at the 10/30 ground glass joint of the stOpcock (C), with enough high vacuum grease and a rotary motion to get a continuous film of grease on the entire ground II . ~ . .. v—unnw . glass surface. This would ensure a leakproof con- nection. The freezing mixture (acetone + dry ice), contained r t in a Dewar flask, was placed around the sample tube to freeze the sample and prevent escape of moisture during evacuation of the apparatus. The sample tube was connected to the manometer by adjusting the stopcock (C). Both sides of the manometer were connected to the vacuum pump by ad— justing the stopcock (B). The vacuum pump was started and the apparatus was evacuated for suffi- cient time to get a uniformly low vacuum, corres- ponding to a pressure of 100 u or lower. The mano- meter was brought into operation by adjusting the stOpcock (B) in a way so that the sample side was sealed off from the vacuum pump, while the other side was being continuously evacuated. The freezing mixture was removed from around the sample tube (D). The sample was allowed to equili— brate to a constant vapor pressure; either to the 23 room temperature in air, or to a predetermined temperature in a constant temperature water bath placed around the sample tube. The temperature of equilibration was recorded. The vapor pressure of the sample was recorded every 5 minutes till two successive readings agreed within 0.2 mm, which was regarded as the equilibrium condition. If a vapor pressure reading at another temperature was desired, the water bath around the sample tube (D) was replaced with another water bath at the new temperature. The vapor pressure of the sample was recorded as before till a new equilibrium was established. The water bath around the sample tube was replaced with the freezing mixture. The water vapor in the vapor space above the sample recondensed, leaving a residual pressure in the manometer due to the desorbed gases during equilibration. The residual vapor pressure was recorded and subtracted from the vapor pressure reading of the sample. The water activity of the sample was calculated by dividing the corrected vapor pressure value of the sample by the vapor pressure of pure water at the temperature of equilibration (converted to mm of Apeizon B oil). 10. 24 Steps 5, 7 and 8 were repeated in some experiments to study the effect of repeated use of a sample through freezing-equilibration cycles on water activity. The freezing mixture was removed from around the sample tube (D). The stOpcock (B) was returned to the original position, connecting both sides of the manometer to the vacuum pump. The vacuum pump was stOpped, and the vacuum in the apparatus was re- leased through the stOpcock (C), by rotating it several times to completely release the vacuum in the sample tube which could not be connected directly to the atmosphere. This limitation of the apparatus has been recognized, and a modified de- sign has been recommended at the end of the chapter on results and discussion. The vacuum in the ap- paratus should be released slowly to prevent the dusting over of the sample into the sample side of the manometer under the fast air current which rushes into the sample tube. The sample tube could then be removed from the apparatus, and the apparatus could be used for another determination. RESULTS AND DISCUSS ION Standardization of the Method Taylor (1961) recommended a 1 gram sample for homo- geneous materials and a 5 gram sample for heterogeneous materials. Spray dried nonfat dry milk is a homogeneous material. A 1 gram sample was, therefore, used for these studies except where the effect of sample size was eval- uated. The results of the process variables studied to standardize the method are discussed in the following pages. Time for Evacuation The objective of evacuating the apparatus was to attain a negligible reference pressure in the manometer to measure the water vapor pressure of the sample. Makower and Myers (1943) recommended 100 u or lower, and Taylor (1961) recommended 10 u pressure in the apparatus after evacuation, to be low enough. This pressure was assumed to be negligible without introducing a Significant error in the results. It was possible to evacuate the apparatus used for these studies to 50 u in one minute. There was very little change in the pressure after one minute. One 25 26 minute evacuation was, therefore, adopted as standard for these studies. Time for Freezing The object of freezing the sample was to prevent evaporative loss of moisture during subsequent evacuation of the apparatus (Taylor, 1961). The time required for freezing is a function of the thermal diffusivity and thick— ness of the sample. The time for freezing was varied from 1 to 15 minutes, and the samples were allowed to equili- brate to the room temperature in air. The results are pre- sented in Table l. The freezing did not have a consistent effect on water activity, equilibration time, and pressure due to de- sorbed gases. The inconsistency in the results was probably due to a difference in the temperatures to which the samples were equilibrated. Water activity is temperature dependent, and the effect of temperature might have been more than the effect of freezing time. Temperature of Equilibration The following three trials were conducted to eval- uate the effect of temperature on water activity, and to find out the optimum time for freezing and evacuation: l. 1 minute freezing and 1 minute evacuation. 2. 2 minutes freezing and 2 minutes evacuation. 3. 2 minutes freezing and 5 minutes evacuation. 27 hflfiJs. I) o.m mm emH.o m.m~ ma m.m mm vum.o m.m~ oa o.m ov Nam.o «.mm m m.m mm , HH~.o m.mm v n.m mm Hmm.o o.m~ m m.n oe mam.o v.mm N m.m mm mmm.o m.mm a Aaao no see Ammuscflsc loo immuscasv mmmmm GOQHOmmU mEHu COflHMHQHHHSUO OH USU whammmum COHMMHnHHHDUm wufl>fluum kuMS MO muduMHmQEmB OEHH mcwnmmuh .memEMm xHHE map pounce co momma omnuommp on map whammmum cam :meflu coflumunwawsgm .wufl>fluow Hmum3 so mfifiu mcfiummum mo pummmmll.a manna 28 The samples were equilibrated to 25 C, by placing a constant temperature water bath around the sample tube. The temperature of the water bath was controlled manually within 0.1 C. After the equilibrium vapor pressure at 25 C was recorded, the 25 C water bath was replaced with a 30 C water bath. The equilibrium vapor pressure at 30 C was also recorded. The pressure due to the desorbed gases was recorded by replacing the water bath with the freezing mixture. The water activity of the sample was calculated at both temperatures. The results are presented in Table 2. Table 2.--Effect of the temperature of equilibration on water activity, equilibration time, and pressure due to desorbed gases on nonfat dry milk samples. Trial Water act1Vity Equilibration Pressure due to time desorbed gases Number 25 C 30 C (minutes) (mm of oil) 0.198 0.211 25 8.2 0.198 0.214 30 9.6 0.197 0.208 30 7.7 The water activity at 30 C was higher than that at 25 C by about 0.016, as expected according to the tempera- ture dependence of moisture sorption isotherms (Rockland, 1969). Taylor (1961) regarded the temperature effect from 18 to 21 C to be negligible for the data collected for pre— paring the desorption isotherms for dehydrated foods. Taylor's assumption could be accepted for a routine work 29 where small differences in the water activity values can be ignored. The error should be eliminated in research work for a more accurate analysis of the experimental data. Ad0ption of a single reference temperature for equilibra- tion, reduced the calculation work as well; as only one vapor pressure value of pure water was required to calcu- late the water activity of the samples. A comparison of the water activity values obtained in the three trials confirmed that 1 minute freezing and 1 minute evacuation was adequate to get reproducible re- sults. No consistent effect was observed on equilibration time and pressure due to desorbed gases. Air Versus Water Equilibration Originally, the samples were allowed to equilibrate to the room temperature in air. When the temperature de- pendence of water activity was found to be significant, a constant temperature water bath was used for equilibration of the samples. A comparison of Tables 1 and 2 revealed that the equilibration time was reduced by about 5-10 minutes, when a water bath was used for equilibration of the samples. The reduction in equilibration time was due to a faster rate of heat transfer from water to the sample, than from air to the sample. Equilibration in air caused freezing of moisture on the sample tube, which increased the resis- tance to heat transfer from air to the sample. The saving 30 in time, however, was not as important a factor as the control of temperature of equilibration. Sample Size Theoretically, water activity is a quality para- meter and should be independent Of sample Size; provided a representative sample is secured. Makower and Myers (1943) used approximately 15 gram samples. Taylor (1961) recommended a 1 gram sample for homogeneous materials, and a 5 gram sample for heterogeneous materials. Spray dried nonfat dry milk is a very homogeneous material, and a 1 gram sample Should be adequate. To find out the Optimum sample size for spray dried nonfat dry milk, the sample size was varied from 1 to 4 grams, while other eXperimental variables were held constant. The results are presented in Table 3. Table 3.--Effect Of sample size on water activity, equili- bration time and pressure due to desorbed gases on nonfat dry milk samples. 3? 'V; k ‘n- ff Sample Water Activity Equilibration Pressure due to Weight time desorbed gases (grams) 25 C 30 C (minutes) (mm Of Oil) 1 0.216 0.225 25 6.5 2 0.210 0.228 30 9.0 3 0.215 0.230 45 11.5 4 0.213 0.232 45 15.0 31 The sample size did not Show a consistent effect on water activity at 25 C, but the water activity at 30 C showed a slight increase with an increase in the sample size. The increase was, however, too small to be of any practical significance. A 1 gram sample was, therefore, considered adequate for the water activity measurement Of Spray dried nonfat dry milk. This confirmed the recom- mendation Of Taylor (1961). The equilibration time increased with an increase in the sample Size. This was expected since a larger sample would require more heat to be transferred to reach equilibrium; thereby requiring a longer time. The residual pressure in the manometer also in- creased with an increase in the sample size. The increase in pressure was proportional to the sample size. This provided a clear evidence that the residual pressure in the manometer was due to gases desorbed from the sample, and not due to leakage Of air into the manometer during equilibration. Sample Tube Capacity During equilibration, Some moisture evaporates from the sample to fill the vacuum space above the sample. Taylor (1961) recommended that the volume of the sample Side of the manometer should be kept to a minimum, if small samples are used, to minimize any possible reduction in water activity due ‘tO evaporation Of moisture from the %; 32 sample. This recommendation was kept in mind while design- ing the present apparatus. It was, however, considered important to evaluate the effect Of increasing the volume of the sample side of the manometer on water activity of the samples. Three sample tubes of approximately 5ml, 10ml, and 15ml volume were used, while the sample size and other process variables were held constant according to the pro- cedure standardized so far. The results are presented in Table 4. Table 4.--Effect Of the volume Of the sample tube on water activity Of nonfat dry milk samples. Water activity at 30 C Sample number Sml tube 10ml tube 15ml tube 1 0.205 0.198 0.205 2 0.207 0.211 0.212 3 0.212 0.207 0.208 4 0.213 0.214 0.214 The volume Of the sample tube did not have a con- sistent or significant effect on the water activity Of the samples. The results were within the normal variability of the method observed so far. It was concluded that the evaporative loss of moisture during equilibration, to fill the vapor space above the sample, was negligible to affect the water activity of the sample. 33 Repeated Use Of a Sample During the preliminary trials on the apparatus, a consistent decrease in water activity was Observed when a sample once analyzed was passed through another freezing- equilibration cycle for a second determination Of water activity. This effect was investigated more rigorously by passing the samples through a number Of cycles without re- leasing the vacuum in the apparatus. The water activity of the samples was calculated for each cycle. The results are presented in Table 5. Table 5.--Effect of the repeated use Of a sample through freezing-equilibration cycles on the water activity Of nonfat dry milk samples. Water activity at 30 C Sample Number lcycle 2cycle 3cycle 4cycle Scycle 6cyc1e 1 0.211 0.208 0.209 0.208 0.209 0.209 2 0.214 0.209 0.208 0.206 3 0.225 0.217 0.219 4 0.228 0.221 0.220 5 0.230 0.224 0.220 6 0.232 0.227 0.226 A small but consistent decrease in water activity was Observed in the second cycle. Very little change was Observed in subsequent cycles. The reduction in water activity was probably due to an increase in the water binding properties of the samples due to the freezing-equilibration cycles. It was 34 possible that the sites vacated by the desorbed gases were taken up by the water molecules, thereby reducing the water activity of the samples. This view was supported by the fact that the pressure due to desorbed gases also in- creased with each cycle (Table 6). The pressure due to desorbed gases was maximum for the first cycle, which could explain the maximum decrease in water activity in the second cycle. The increase in the pressure due to desorbed gases decreased with each cycle. This could ex- plain the relatively small decrease in water activity after the second cycle. Table 6.--Effect of the repeated use of a sample through freezing-equilibration cycles on pressure due to desorbed gases Of nonfat dry milk samples. Pressure due to desorbed gases (mm Of Oil) Sample Number lcycle 2cycle 3cycle 4cycle Scycle 6cycle 1 8.2 10.5 12.0 14.0 14.8 16.0 2 9.6 13.0 15.0 17.0 3 6.5 8.5 9.5 4 9.0 11.0 12.0 5 11.5 14.5 16.5 6 15.0 19.0 22.0 These Observations suggest a further investigation into the effect Of cycling on the water activity Of foods which have different porosity and tissue structures, like dehydrated fruits and vegetables. 35 Sealing Both Sides of the Manometer After the initial evacuation Of the apparatus, Makower and Myers (1943) and Taylor (1961) kept the refer- ence side of the manometer evacuated continuously during equilibration of the samples. The Objective was to main- tain a uniformly low pressure in the reference arm of the manometer tO measure the vapor pressure of the sample accurately. This procedure was followed for the data re- ported so far. Because the apparatus was evacuated to practically zero pressure, it was thought that the reference side of the manometer could also be sealed Off from the vacuum pump, without introducing a significant error in the re- sults. This procedure would eliminate the need to keep the vacuum pump running during equilibration of the samples. Some increase in pressure in the reference arm Of the mano- meter was inevitable during equilibration of the samples with the rise in level Of the Oil in the manometer. This increase in pressure could be negligible in view of the' very low pressure created in the apparatus. It was con- sidered desirable to investigate the possible application Of the modified procedure. The procedure was modified to adjust the stOpcock (B), after the initial evacuation of the apparatus, in a way to bring the manometer into Operation and seal Off the 36 reference side of the manometer from the vacuum pump. The sample was allowed to equilibrate to 30 C in a water bath, and the equilibrium vapor pressure was recorded. The stOp- cock (B) was readjusted tO connect the reference side of the manometer to the vacuum pump as in the original pro- cedure. The vapor pressure increased and a new equilibrium was quickly attained. The new equilibrium vapor pressure was also recorded. The standard procedure was followed beyond this point and the water activity Of the samples was calculated for the two conditions. The evacuation time was varied from 1 to 5 minutes, to see if longer evacuation time had any effect on the results. The results are pre- sented in Table 7. Table 7.--Effect of sealing both sides Of the manometer on the water activity of nonfat dry milk samples. Water activity at 30 C Evacuation (M33228) :23: 222:: “22.33“ 1 0.222 0.199 0.023 1 0.242 0.219 0.023 1 0.224 0.202 0.022 1 0.241 0.220 0.021 2 0.224 0.211 0.013 2 0.244 0.228 0.016 2 0.245 0.232 0.013 3 0.223 0.209 0.014 4 0.231 0.217 0.014 5 0.246 0.237 0.009 37 The water activity of the sample was always higher when the reference arm Of the manometer was connected to the vacuum pump, compared to the water activity when both sides of the manometer were sealed. The difference was quite significant, but consistent for the same evacuation time of 1 minute. The difference decreased for longer evacuation times due, possibly, to a slightly better evac- uation of the apparatus. In view of the large differences Observed in the water activity of the samples under the two conditions, the modified method cannot be recommended for use. The method, however, could be used with a reference curve to correct the Observed data. There was the additional risk that any leakage in the reference side of the manometer would go undetected. The method was not used for this research work. Reproducibility of the Method A sample of spray dried nonfat dry milk was col- lected in a large airtight glass bottle. The bottle was filled with the sample up to the neck, to minimize the effect of adsorption of moisture from the atmOSphere during repeated sampling for analysis. Ten replicate samples were analyzed in the apparatus and the water activity data was statistically analyzed for mean and standard deviation. The data is presented in Table 8. 38 Table 8.--Water activity data Of spray dried nonfat dry milk for statistical analysis. ~évey - +~.-.' Replicate number Water activity at 30 C 1 0.204 2 0.209 3 0.205 4 0.209 5 0.199 5 0.204 7 0.204 8 0.206 9 0.198 10 0.202 Notes: Mean = 0.204. Standard deviation = 0.00365. 39 Accuracy of the Method The samples of three size fractions Of Spray dried nonfat dry milk were equilibrated to constant weight in controlled humidity environments. The samples were analyzed for water activity. The results presented in this section are averages Of duplicate analysis on each sample. The water activity Of the samples equilibrated over Phosphorus pentoxide (water activity = 0) are presented in Table 9. The water activity Of the samples equilibrated Table 9.--Water activity of nonfat dry milk samples equili- brated over Phosphorus pentoxide. Nonfat dry milk fraction Water activity at 30 C 44 to 63 u 0.008 63 to 88 u 0.011 88 tO 125 p 0.010 over Phosphorus pentoxide was expected to be near zero. The small values found were most probably due to adsorption Of moisture from the atmosphere when the sample was trans- ferred to the sample tube, because the desiccated samples of dry milk are very hygroscopic. The water activity Of samples equilibrated over saturated salt solutions under vacuum and at room tempera- ture are presented in Table 10. The water activity of the saturated salt solutions was also found out for comparison with the literature values (Table 10). «w. 40 Table 10.--Water activity Of nonfat dry milk samples equilibrated over saturated salt solutions under vacuum and at room temperature. Water activity at 30 C Saturated salt . . . . Saturated salt solution Nonfat dry milk fractions solution Literature Experimental 44-63 u 63—88 0 88-125 u LiCl 0.112 0.109 0.162 0.163 0.160 KAc 0.220 0.213 0.268 0.260 0.279 MgCl2 0.324 0.319 0.366 0.370 0.373 The water activity of all the three saturated salt solutions was slightly lower than the values reported by Wink and Sears (1950) by differential vacuum manometry with an accuracy of $0.002. The difference from the literature values could be due to experimental errors. The literature values have been used for comparing the water activity of the samples equilibrated over saturated salt solutions. The water activity of the dry milk samples was much higher than the water activity Of the corresponding satur— ated salt solutions. Such a large difference could have been due to adsorption Of moisture from the atmospheric air when the vacuum was released in the desiccators. The rela- tive humidity Of the air was about 40-50 percent, which is higher than the equilibrium relative humidity Of all the samples tested. 41 The samples were, therefore, allowed to equilibrate again at atmospheric pressure. The samples were analyzed again for water activity. The results are presented in Table 11. Table ll.--Water activity of nonfat dry milk samples equilibrated over saturated salt solutions at atmospheric pressure and at room temperature. Water activity at 30 C Saturated Salt. Saturated salt solution Nonfat dry milk fractions Solution (Literature) 44-63 p 63-88 u 88-125 u LiCl 0.112 0.144 0.148 0.159 KAc 0.220 0.259 0.249 0.249 MgCl2 0.324 0.356 0.354 0.353 The water activity Of the dry milk samples was lower than the water activity Of the same samples under vacuum equilibration (Table 10). But the values were still much higher than the corresponding values for the saturated salt solutions. At this stage it was realized that the samples were equilibrated at the room temperature (25-27 C), but the water activity was measured at 30 C. This introduced a temperature effect which could be quite significant to affect the results. Generally the samples are equilibrated to the room temperature for the water activity measurements: so that the temperature of equilibration and the 42 temperature of measurement are the same. In this study a water bath was introduced to controltflmatemperature Of equilibration. The temperature error was inadvertently introduced. A comparison Of Tables 10 and 11 Showed that the particle size had no effect on the water activity of the samples. The study of particle size as a variable was discontinued at this stage. The samples were further equilibrated in an incu- bator maintained at 30$0.5 C, and were analyzed for water activity. The results are presented in Table 12. The dry milk samples showed a much better agreement with the sat- urated salt solutions. The water activity of the dry milk samples was slightly higher at lower water activity of the saturated salt solutions, and vice versa. This could have been due to incomplete equilibration of the samples as Ob- served by Elvanides and Markakis (1971). Table 12.--Water activity of nonfat dry milk samples equili— brated over saturated salt solutions at atmos- pheric pressure and at 30 C. Water activity at 30 C Saturated salt solution Saturated salt Nonfat dry solution milk (Literature) LiCl 0.112 0.124 KAc 0.220 0.231 MgCl 0.324 0.314 43 A final check was made on the accuracy and the reproducibility of the method by equilibrating a new sample Of nonfat dry milk over a saturated salt solution of Potas- sium acetate (KAc) at atmospheric pressure and at 30$ 0.5 C. Potassium acetate was selected because its water activity is very near to the water activity Of the commercial sam- ples Of nonfat dry milk. This gave an evaluation of the method nearest to the possible application Of the method. Ten replicates Of the sample were analyzed, and the data was statistically analyzed for mean and standard deviation (Table 13). The mean (0.224) was very close to the literature value for Potassium acetate (0.220), but was little higher than the experimental value (0.213). The standard devia- tion was about half Of the value found earlier (Table 8). The method has a high reproducibility to the third decimal place and is accurate to within 0.01 water activity. Rate Of Equilibration Taylor (1961) recommended grinding of the food samples to increase the rate Of equilibration. The effect of particle size on the rate Of equilibration was evaluated by using different size fractions of regular Spray dried and instant nonfat dry milk samples. The time required_to reach equilibrium as a function Of particle size fraction of regular Spray dried nonfat dry milk, showed a small but Table 13.—-Water activity data Of Spray dried nonfat dry milk sample equilibrated over a saturated salt solution of Potassium acetate at atmospheric pressure and at 30$0.5 C for statistical analysis. 44 Replicate number Water activity at 30 C 9 10 0.223 0.225 0.218 0.227 0.223 0.223 0.226 0.226 0.225 0.227 Notes: Mean = 0.224. Standard deviation 0.0018. .A'_.T&_ ‘1 ‘ {J‘- A ‘— "W ._.-,-.-__ v "7 n.- 45 perceptible increase with a decrease in the particle size fraction (Table 14). Table l4.--Effect Of particle Size fraction of regular spray dried nonfat dry milk on the time required to reach equilibrium. Time required to reach equilibrium (minutes) Saturated salt solution 44-63 p 63—88 u 88—125 u fraction fraction fraction Lithium chloride 26 25 23 Potassium acetate 23 25 22 Magnesium chloride 27 25 25 The average time to reach equilibrium was 25 min— utes. The plots Of vapor pressure versus time, during equilibration, did not Show any clear trend as a function Of particle size fraction. The range of particle size fraction studied was considered very narrow. The experi- ment was extended to study the effect of large agglomerate size fractions of instant nonfat dry milk. The results on water activity and the time required to reach equilibrium are presented in Table 15. The water activity Of the samples did not show a clear trend as a function of agglomerate size fraction. The differences in the water activity of different frac- tions were possibly due to adsorption of moisture during 17 _‘Q 46 Table lS.--Effect Of agglomerate size fraction of instant nonfat dry milk on water activity and time re- quired to reach equilibrium. Agglomerate size Water Time required to reach fraction activity equilibrium (minutes) < 125 u 0.243 14 125 — 177 u 0.229 14 177 - 250 u 0.223 14 1 250 - 500 p 0.227 10 71 500 - 707 u 0.226 11 : 707 - 1000 u 0.234 10 g 1000 - 1410 p 0.271 11 (' sieving. The differences were more for the fractions which were smaller, showing more adsorption Of moisture. The time required to reach equilibrium again showed a small but perceptible increase as the agglomerate size fraction decreased. The average time required to reach equilibrium was only about 12 minutes compared to 25 min- utes for the regular Spray dried nonfat dry milk. The plots of vapor pressure as a function of time during equilibration showed an increase in the time lag before the pressure started rising in the manometer, as the agglomerate size fraction decreased. The increase in time lag was probably due to cooling of the samples to lower temperatures because the smaller particles pack more closely and reduce the resistance to heat transfer. This raised the possibility that samples with very large parti— cles or chunks, like dehydrated fruits and vegetables, 47 might not be cooled to low enough temperature to prevent moisture loss during evacuation. NO reduction in water activity was Observed for the instant dry milk samples with larger agglomerate size fractions. It is possible that the sample tube itself works as a moisture trap so that any water vapor trying tO escape freezes on the walls Of the sample tube. This factor, however, needs further investi- gation before any clear cut eXplanation can be provided. The time lag during equilibration has been illus- trated in Figure 2. The plots represent the average fit for duplicate runs on each agglomerate size fraction, to show the trend Of increase in time lag with a decrease in the agglomerate Size fraction. .The equilibration data was further analyzed for the time lag on the analogy of the time lag in the heating curves for the food products. The following equation was used: 10g(Pe-P) = JT/t + log L(Pe-PO) where Pe = product vapor pressure at equilibrium. P = product vapor pressure at any time t. P0 = product vapor pressure at the start of equilibration. T = constant describing the rate of change in vapor pressure P. L = constant describing the lag in time required for change in vapor pressure to become logarithmic. 48 .ucmumcwllxawa who unwaoc mo :OAuomum onwm ounuufionmu mo c0auocau a no :Oauouawawsuo mo muumnu.~ «Manda mouscwe Icouumum :ofiuuhpwafizam Houmm mafia 2 S m o v a a a W q 1 1 d d mfiv .k 4 m2 4.2 S a A: .53 .m N a 1 SN 18m. .1. 1 8m 4.2 .m a SS :82 .n a 82-23 .H l C v {to 30 mm ~axnssaxd IOdBA l 0 w .1 OS 49 The plots of Pe-P versus t on a semilogarithmic graph paper very closely resembled the heating curves for food products. The lag factor L was calculated for each curve using the following equation: L = Pe-Pa/Pe-PO where Pa is the pressure on the y axis where the Pe-Po versus t curve intersects it at zero time. The L values showed an increase with a decrease in the agglomerate size fraction (Table 16); supporting the earlier conclusion that the time lag in equilibration was, possibly, due to faster cooling of the samples with smaller agglomerates. Table 16.--L value as a function Of the agglomerate size fraction of instant nonfat dry milk. Agglomerate Average agglomerate L value size fraction diameter (u) 1 2 average < 125 u 62.5 12.96 22.66 17.81 125 - 177 u 151.0 11.73 9.34 10.54 177 - 250 u 213.5 3.37 5.88 4.63 250 - 500 u 375.0 4.26 3.53 3.89 500 - 707 u 603.5 5.14 3.31 4.23 707 - 1000 u 853.5 5.07 3.07 4.07 1000 - 1410 u 1205.0 2.16 2.98 2.57 To investigate the effect of agglomerate size fraction on L value, the L value was plotted against the average agglomerate diameter (Figure 3). The curve Showed 50 .ucmumcflnlxafie mun ummcoc mo HmumEMHO mumumEOHmmm mmmum>m may no cOHuOGSM o no osHm> q||.m musmam .1 uhmmemwv mumnoeoawwm owgmg. coma cow oov L I, F 4 u 1 q 4 « 51 a very good fit, but there was a break in the curve at about 200 u. L value increased very rapidly below 200 u, but it was relatively independent of agglomerate size above 200 u. The same trend could be seen in Figure 2. It is difficult to provide an explanation for this peculiar be- havior without further investigation. Recommended Design and Procedure The following modifications in the design of the apparatus are suggested to further improve the apparatus: 1. The ground glass stopcock (C) is not required. It should be removed from the apparatus and the sample arm Of the manometer should be directly connected to the ground glass joint which accepts the sample tube (D). 2. The 10/30 ground glass Opening of the sample tube (D) is too small to allow quick transfer of the samples. This sample tube cannot be used for analyzing foods in the form Of large chunks, like dehydratedfruits and vegetables. A wider ground glass joint Should be used. 3. To compensate for the first modification, a device to release the vacuum in the apparatus Should be provided in the vacuum line. 52 The following standard procedure is recommended to determine the water activity of nonfat dry milk samples: 1. 10. 11. 12. Transfer about 1 gram sample to the sample tube. Connect the sample tube to the apparatus. Freeze the sample for 1 minute. Evacuate the apparatus for 1 minute. Bring the manometer in Operation. Replace the freezing mixture with a constant temperature water bath or allow to equilibrate in air. Record the equilibrium vapor pressure of the sample. Replace the water bath with the freezing mixture. Record the residual pressure in the manometer. Remove the freezing mixture and connect both sides Of the manometer to the vacuum pump. StOp the vacuum pump and release the vacuum in the apparatus. Subtract the residual pressure from the equilibrium vapor pressure of the sample, and divide by the vapor pressure of pure water at the temperature Of equilibration; to get the water activity of the sample. SUMMARY AND CONCLUSIONS A simple apparatus was designed to evaluate the water activity of foods. The procedure to measure the water vapor pressure of foods was standardized, using Spray dried nonfat dry milk as a model system. A 1 minute evacuation time was found adequate to evacuate the apparatus to a consistently low pressure of about 50 u. An increase in evacuation time did not affect the final vacuum attained and the water activity of the sample. A 1 minute freezing was found enough to prevent evaporative loss of moisture during the subsequent evacua- tion of the apparatus. An increase in freezing time did not affect the water activity of the sample and the time required for equilibration. The water activity showed a positive correlation with the temperature Of equilibration. The water activity at 30 C was higher by about 0.016 than the water activity at 25 C. The time required to reach equilibrium was de— creased by about 5-10 minutes, when a water bath was used in place of air for equilibration. 53 54 A small increase in water activity at 30 C was Observed with an increase in the sample size. The time required to reach equilibrium increased with an increase in the sample Size. The pressure due to desorbed gases increased fairly proportionately with the sample size. The volume of the sample tube did not have any effect on the water activity of the samples. The water activity of the samples decreased, when the samples were subjected to repeated freezing- equilibration cycles. The main decrease was in the second cycle, about 0.006. The water activity was observed to be higher by about 0.022, when the reference side of the manometer was being continuously evacuated; than the water activity when the reference side was sealed Off from the vacuum pump. The difference, however, was fairly consistent and a ref- erence curve could be develOped to correct the results Obtained by the modified method. The method had a very good reproducibility. The standard deviation for a commercial sample was 0.00365 with a mean Of 0.204. The standard deviation Of a sample equilibrated over saturated salt solution of Potassium acetate was 0.0018 with a mean of 0.224. The method was found to be accurate within $0.01 water activity. Vacuum and temperature were found to affect the results very significantly. The small 55 variations from the standard values were, most probably, due to incomplete equilibration of the samples. The particle size did not have a significant ef- fect on water activity and rate of equilibration. An in- crease was Observed in the time lag,before the vapor pres- sure started to increase rapidly, with a decrease in 5 particle size. This was thought tO be due to a better freezing of the samples with smaller particle size which allows better and closer packing of the sample. The time required for equilibration was only about 12 minutes for instant compared to about 25 minutes for the regular samples. REFERENCES REFERENCES Acker, L. 1969. Water activity and enzyme activity. Food Technol. 23:1257. Bone, D. P. 1969. Water activity--its chemistry and applications. Food Product DevelOpment. 3(5):81. Elvanides, S. N., and P. Markakis. 1971. Preliminary results on rapid water activity measurement. (Unpublished.) Gibson, R. E., and L. H. Adams. 1933. Changes Of chemi- cal potential in concentrated solutions Of cer— tain salts. J. Am. Chem. Soc. 55:2679. Karel, M., and T. P. Labuza. 1968. Nonenzymatic browning in model systems containing sucrose. J. Agr. Food Chem. 16:717. Karel, M., and J. T. R. Nickerson. 1964. Effect of rela- tive humidity, air and vacuum on browning Of de- hydrated orange juice. Food Technol. 18:1214. Labuza, T. P. 1968. Sorption phenomenon in foods. Food Technol. 22:263. Labuza, T. P., H. Tsyuki, and M4 Karel. 1969. Kinetics Of linoleate oxidation in model systems. J. Am. Oil Chem. Soc. 46:409. Labuza, T. P., L. Mcnally, D. Gallagher, J. Hawkes, and F. Hurtado. 1972,a. Stability Of intermediate moisture foods. 1. Lipid oxidation. J. Food Sci. 31:154. Labuza, T. P., S. Cassil, and A. J. Sinskey. l972,b. Stability Of intermediate moisture foods. 2. MicrobiOlOgy. J. Food Sci. 31:160. 56 57 Makower, B., and S. Myers. 1943. A new method for the determination of moisture in dehydrated vegetables. Proc. Inst. Food Technology. 4th conference. p. 156. Mossel, D. A. A., and H. J. L. Kirijk. 1955. A new simple technique for the direct determination of equilibrium relative humidity of foods. Food Research. 223415. Rockland, L. B. 1957. A new treatment of hygroscopic equilibria: application to walnuts and other foods. Food Research. 22:604. Rockland, L. B. 1969. Water activity and storage stability. Food Technol. 22:1241. Salwin, H. 1959. Defining minimum moisture contents for dehydrated foods. Food Technol. 223594. Salwin, H. 1963. Moisture levels required for stability in dehydrated foods. Food Technol. 22:34. Scott, W. J. 1957. Water relations of food spoilage organisms. Advan. Food Res. 'lz83. Stitt, F. 1958. Moisture equilibrium and the determina- tion of water content Of dehydrated foods. Funda- mental ASpectS Of the Dehydration of Foodstuffs. Society of’Chemical Industries, London, England. Taylor, A. A. 1961. Determination of moisture equilibria in dehydrated foods. Food Technol. 22:536. Wink, W. A., and G. R. Sears. 1950. Instrumentation studies LVII--Equilibrium relative humidity above saturated salt solutions at various temperatures. TAPPI. 22(9):96A.