--—_—_v ___ _ “ A GAS mmmmeawmc swnv OF 1% FLAVOR DEWEEEQEATEQN {N {-MGH.TEMEIEa/amasgemaamsm Ewen?» svgmg 5‘4th Thesis for {‘59 Degree of M. 5. MICHIGAN STATE UNIVERSITY James R. Kirk 1966 \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ 3 1293 103 W ‘ MSU LIBRARIES RETURNING MATERiALs: Piace in BooE arop to remove this checkout from your record. FINES w111 be charged if book is returned after the date stamped be10w. _.__- ABSTRACT A GAS CHROMATOGRAPHIC STUDY OF THE FLAVOR DETERIORATION IN HIGH-TEMPERATURE-SHORT-TIME FLUID STERILE MILK by James R. Kirk A study has been made of the flavor deterioration that takes place in high-temperature-short-time (HTST) fluid sterile milk during storage at 4.40, 22°, and 36°C. for three months. Fluid milk was sterilized in the commercial DeLaval Vacu—Therm-Instant-Sterilizer unit at a temperature of 140.50C. 1 10C. for approximately 4.0 seconds; the milk samples were then collected in previously sterilized glass containers in an aseptic atmosphere. Body defects that developed in the fluid sterile milk during storage were cream layer formation, age-thickening. and sedimentation. All sterile milk samples showed the pre- sence of sedimentation by one month of storage. Age-thickening, an incipient sign of gelation, was observed in the 220 and 36°C. storage samples by two months of storage, but was not evident in the 4.4OC. storage samples until three months. Cream layer formation was found to occur after only one month in the storage samples held at 220 and 36°C. This body defect was not evident in the 4.4OC. samples after three months storage. James R. Kirk Organoleptic evaluation, which was carried out during the storage of the HTST fluid sterile milk, indicated that stale flavor was present in the milktafter two months of storage at 22°C. This sensory evaluation also indicated that the rate of product staling was a function of the temperature at which the product was stored, since fluid sterile milk was judged slightly stale after only one month of storage at 36°C. Separation of the flavor components, which had been isolated from the HTST fluid sterile milk by low—temperature, reduced-pressure distillation, was accomplished by gas chroma- tography using a 1/4 inch Q.D. x 6 foot packed column and a dual hydrogen flame detector. The flavor components that were tentatively identified in the sterile milk control samples (zero time storage) by retention time and functional group analysis data were: ethanal, propanal, furfural, ethanol, butanol, acetone, and 2—butanone. Storage samples yielded in addition to the aforementioned: 2—pentanone, 2-heptanone. butanal, hexanal, and heptanal. Also, it was observed that the fluid sterile milk samples that had been judged stale by organoleptic evaluation exhibited a 30-50% increase in acetone concentration, as well as a definite increase in all of the tentatively identified carbonyl compounds. Confirmation of some of the flavor compounds tenta- tively identified by the gas chromatography was made by thin- layer chromatographic separation of their 2,4 dinitrophenylhy- drazones. These hydrazone derivatives were first separated on a Magnesia-Celite 545 column according to homologous James R. Kirk classes prior to thin-layer separation. MBthyl ketones, saturated alkanals. and 2,4 dienals were the three classes identified on the absorption column; however, confirmatory TLC could only be carried out on the methyl ketones and saturated alkanals. These data confirmed the presence of ethanal, acetone, and 2—butanone in the control samples, while ethanal, butanal, hexanal, acetone, 2—butanone, and 2-heptanone were isolated from the three months 22°C. storage samples. Analysis of the data collected from both gas and thin- layer chromatography indicates that the stale flavor is probably due to the synergistic interaction of these tenta- tively identified compounds with those developing during product storage playing a more definite role in the stale flavor. A GAS CHROMATOGRAPHIC STUDY OF THE FLAVOR DETERIORATION IN HIGH-TEMPERATURE-SHORT-TIME FLUID STERILE MILK BY James R. Kirk A THESIS Submitted to Michigan State university in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Food Science 1966 ACKNOWLEDGEMENTS My sincere thanks are extended to Dr. C. M. Stine for his counSel and guidance throughout this study and in the preparation of this manuscript. Appreciation is also extended to Dr. J. R. Brunner, Dr. H. A. Lillevik and Dr. T. I. Hedrick, who reviewed this manuscript. The author is also grateful to the DeLaval Separator Company, who partially financed this research. ii INTRODUCTION . . . REVIEW OF LITERATURE TABLE OF CONTENTS Flavor Defects Associated with Sterile Dairy Products Methods of Sampling Flavor Compounds Means of Identification of Isolated Components Flavor Components of Milk and Milk Products Origin of Flavor Compounds EXPERIMENTAL PROCEDURES Sample Collection Techniques Vblatile Recovery Gas Chromatography Sampling of the Flavor Isolate for Analysis Thin-Layer Chromatography Bacterial and Organoleptic analyses of Storage Sterile Milk Samples RESULTS AND DISCUSSION . Organoleptic Flavor Evaluation Body Defects Associated with HTST Fluid Sterile Milk During Storage at 4.40, 22°, and 36°C. Parameters Incurred in Flavor Recovery Distillation, Identification of Flavor Compounds Detected in HTST Fluid Sterile Mfllk by Gas Chromatography Flavor Vblatiles Tentatively Identified in HTST Fluid Sterile Hulk by Gas Chromatography and Their Flavor Significance Analyses of 2.4 Dinitrophenylhydrazones Derivatives Sampling, SUMMARY AND CONCLUSIONS. . LITERATURE CITED . and Gas Chromatography iii 30 31 33 35 36 38 39 41 41 45 59 81 91 94 LIST OF TABLES Table Page 1. Chronological review of flavor components isolated from stored dairy products . . . ll 2. Organoleptic evaluation data for HTST fluid sterile milk under storage conditions of 4.4, 22. and 36°C. for a three month time interval (processing date 3/2/66) . . . . . . . . . . . . . . . . 46 3. Organoleptic evaluation data for HTST fluid sterile milk under storage conditions of 4.4, 22, and 36°C. for a three month time interval (processing date 5/20/66) . . . . 46 4. Organoleptic evaluation data for HTST fluid sterile milk under storage conditions of 4.4, 22, and 36°C. for a three month time interval (processing date 6/9/66) . . . . 47 5. Retention times of flavor components detected in HTST fluid sterile milk during a three month storage survey by gas chromatography . . . . . . . . . . . . . . 60 iv Figure 10. 11. LIST OF FIGURES gas chromatograph: of the flavor volatiles from HTST fluid sterile milk control sample . . . . . . . . . . . . . . . . . gas chromatograph of the flavor volatiles from HTST fluid sterile milk after one month storage at 4.4°C. . . . . . . . . gas chromatograph of the flavor volatiles from HTST fluid sterile milk after one month storage at 22°C. . . . . . . . . gas chromatograph_ of the flavor volatiles from HTST fluid stegile milk after one month storage at 36 C. . . . . . . . . gas chromatograph of the flavor volatiles from HTST fluid sterile milk after two months storage at 4.4°C. . . . . . . . gas chromatograph of the flavor volatiles from HTST fluid sterile milk after two months storage at 22°C. . . . . . . . . gas chromatograph of the flavor volatiles from HTST fluid sterile milk after two months storage at 36°C. . . . . . . gas chromatograph of the flavor volatiles from HTST fluid sterile milk after three months storage at 4.4°C. . . . . . . . . gas chromatograph of the flavor volatiles from HTST fluid sterile milk after three months storage at 22°C. . . . . . . . . gas chromatograph of the flavor volatiles from HTST fluid sterile milk after three months storage at 36°C. . . . . . . . in chloroform recovered.from fraction number 1 (gray methyl ketone band) during the column absorption class separation of Page 62 63 64 65 66 67 68 69 70 71 Visible absorption spectra of 2L4 DNP hydraZones 2,4 DNP hydrazones from sterile milk at 36 C. for three months, and control samples . V 88 Figure 12. 13. 14. 15. l6. 17. Page Visible absorption Spectra of 2,4 DNP hydra- zones in chloroform recovered from" fraction number 2 (tan saturated alde- hyde band) during the column absorption class separation of 2,4 DNP hydrazones from sterile milk at 360C. for three months, and control samples . . . . . . . 88 Visible absorption spectra of 2.4 DNP hydra- zones in chloroform.recovered from fraction number 3 (tan saturated aldehyde band) during the column absorption class separation of 2, 4 DNP hydrazones from sterile milk at 360 C. for three months, and control samples . . . . . . . . . . . 88 Separation of 2,4 dinitrophenylhydrazone derivatives of saturated aldehydes from HTST fluid sterile milk control samples by thin-layer chromatography on Carbowax 400 impregnated plates of Kieselguhr G. Mobile phase, petroleum ether loo-1200c. . 89 ,Separation of 2,4 dinitrophenylhydrazones derivatives of saturated aldehydes from HTST fluid sterile milk, held at 22°C. for three months, by thin-layer chroma- tography on Carbowax 400 impregnated plates of Kieselguhr G. Mobile phase, petroleum ether lOO- 120° C . . . . . . . . 89 Separation of 2,4 dinitrophenylhydrazone derivatives of methyl ketones from HTST fluid sterile milk control samples by thin-layer chromatography on Carbowax 400 impregnated plates of Kieselguhr G. Mobile phase, petroleum ether 100- 120°C. . . . . . . . . . . . . . . . 9O Separation of 2, 4 dinitrophenylhydrazone derivatives of methyl ketones from HTST fluid sterile milk, held at 22°C. for three months, by thin-layer chromatography on Carbowax 400 impregnated plates of Kieselguhr G. Mobile phase, petroleum ether loo—120°C. . . . . . . . . . . . . 90 vi INTRODUCTION The importance of milk as an article of food in the human diet existed from the time of recorded history as can be attested by the ancient Sanskrit writings of India. Dairy conditions from these ancient times until 1850 showed little change either in the methods of production or manufacturing: thus each family depended upon privately owned cattle for its supply of milk and dairy products. The process of pasteuri- zation was developed by Pasteur in 1864, and a few years later Jacobi and Soxhlet, realizing the value of pasteuriza- tion as a public health safeguard and a means of increasing the self-life of milk, advocated the pasteurization of milk. By the early 19th century pasteurization was common practice in America and Europe. Although pasteurization is effective in destroying the pathogenic bacteria, and the bulk of the non—pathogens in milk, it does not preclude bacterial spoilage. Thus a new processing method, milk sterilization, was developed as a means of increasing the shelf life of dairy products by elimination of bacterial spoilage. Sterilization has proved effective against bacterial deterioration: however, the re- sultant sterile products have a flavor which is much less desirable than that of normally processed products. Storage of these sterile products for prolonged time periods results in further flavor deterioration characterized as "stale" or "stale-oxidized”. The stale or stale-oxidized flavor defect has been shown to arise in part from the deterioration of the milk lipids. Webb, in 1963, commented on the formidable task of obtaining an acceptable flavor in sterile dairy products. He stressed the fact that both the stale-oxidized, and the cooked-carmelized flavors, that now develop in commercially sterilized products must be eliminated before their markets can be fully exploited. To accomplish the task of maintaining normal product flavor, researchers have concluded that the following three steps must be achieved: a) all compounds contributing to off-flavor must be detected, b) origins of off—flavor compounds must then be elucidated, and c) steps taken to prevent for- mation of these off-flavor compounds. Flavor chemists feel that attainment of these goals ultimately lies in the elucidation of the compounds that con— tribute to the natural falvor of the fresh product. Utilizing this information as a common reference point, research can then be directed to the problem of off—flavor detection and analysis. It is the intention of this research, with the aid of some of this previously secured information on the flavor chemistry of dairy products, to shed some light on the problem of stale—oxidized, or as it is more commonly known storage flavor, which is a common defect of heat-processed dairy products upon storage. The primary concern of this research is the study of the development of off-flavors.in high-temperature, short—time (HTST) processed fluid sterile milk under storage conditions of refrigerated, ambient, and tropic temperatures. REVIEW OF LITERATURE Flavor Defects Associated with Sterile Dairy Products The most common flavor defect associated with fluid sterile milk is the stale or stale-oxidized flavor that usually manifests itself between the second and third month of product storage at 70°F. (Patel et al., 1962). The stale or stale- <3xidized flavor defect, although never fully defined, has been studied by Bassette and Keeney (1960), Nawar et a1. (1963), Bingham (1964), Arnold et a1. (1966), and all have indicated that both the Maillard browning reaction and lipid oxidation may be involved in its formation. Other off-flavors which detract from the flavor of sterile dairy products are the cooked flavor and the sweet- carmel taste, which are present because of the stringent heating conditions needed for product sterilization (Harland and Ashworth, 1945; Gould and Lomoner, 1939; Hutton and Patton, 1952; and Hodge, 1953). Thus, if we are to obtain a product that has any future market value, we must find a way to prevent these un- desirable flavor deteriorations from occurring whether they be in processing or during storage. Sample Collection Techniques Sampling flavor compounds of foods has proven to be a somewhat difficult task because of the relative ease with which artifacts can be incorporated into the sample by the recovery procedure. This is especially true of dairy products, since they easily pick up off—flavors and only small concen- trations of some compounds, in some cases parts per billion, are needed to cause these off—flavors. The isolation of flavor components from milk is even more difficult than most foods, since fresh milk normally has a delicate bland taste and aroma. Therefore, it was necessary to first develop recovery techniques capable of concentrating these flavor components to a point whereby they could be detected by present— day scientific instruments, and at the same time assure the original integrity of the compounds responsible for imparting these flavors. Mabbitt and McKinnin (1963) removed the volatile components of milk by entrainment of the volatiles in a sweep gas and condensed them in a trap cooled by liquid nitrogen. They felt this method caused minimal change in composition of the volatiles. The success of this method depended only on having a pure gas supply in which to entrain the volatiles. Libbey et al. (1963) utilized the technique of steam stripping under reduced pressure for recovering the flavor volatiles assoc1ated with the lipid phase in dairy products of high fat content. This method also utilized a fractional collection technique for recovering the volatiles from the distillate by using various trap temperatures ranging from 00C. to —195.8°c. Winter et a1. (1963) recovered the volatile components of butter by an oxygen-free water emulsification of the butter, followed by a reduced pressure distillation of the emulsion in a falling-film, thin—layer distillation column. This technique had value in that by performing these steps in a medium of nitrogen, atmospheric oxidation of the fat during volatile extraction was eliminated, thus eliminating artifacts due to oxidation. El-Negoumy et a1. (1961) utilized a molecular still and Skellysolve B in isolating and concentrating butter- oil volatiles. Patton (1961) suggested a unique solvent extraction technique for recovering flavor compounds in processed milks. This was accomplished by allowing the protein adsorbed on the fat and that of the fat globule membrane to act an an interface and prevent contact between fat and solv€nt, thereby excluding unwanted fat from the extracting solvent. A reverse-phase, liquid-liquid extraction system for recovering the volatile organic components of citrus juices was developed by Senn (1963). Nawar and Fagerson (1960) employed gas-liquid extraction in.a recycling system for recovery of food volatiles, which permitted recovery of high percentages of the flavor volatiles because of minimal losses to the atmosphere. Bassette et a1. (l962)*developed a means of studying milk volatiles by sampling headspace gas. Here a 2 ml. sample of the milk to be analyzed was added to a serum vial containing 1.2 gm. of sodium sulfate. The vial was then capped with a rubber septum, shaken for 5 minutes, and heated at 600C. for 3 minutes. Sampling of the headspace gas was accomplished by inserting the needle of a gas-tight syringe through the rubber septum and removing the desired volume of sample. This pro— cedure enabled the recovery of volatiles at a concentration as low as 1 ppm. Recently, Morgan and Day (1965) have adapted headspace gas sampling to the on—column trapping system developed by Hornstein and Crowe (1962) using nitrogen as the entrainment gas and liquid nitrogen as the cooling medium for condensing the flavor volatiles on the packed column. This eliminated the need of gas transfer systems and the heating of volatile traps; extremely good reproduction is obtained with this method. Bradley and Stine (1962) devised a method for sampling headspace gas from sealed tin containers. The device forms an air-tight puncture in the lid of the container, from which the desired volume of gas is then removed by inserting the needle of a gas-tight syringe through the rubber septum of the sampling device. MacKay and Berdick (1961) also used a method of sampling headspace gas for analysis of flavor volatiles. Reduced-pressure, low-temperature distillation technique was also developed by Day et a1. (1957) for collect— ing flavor compounds from gamma—irradiated milk. This method utilized the technique of fractional collection of flavor volatiles in wet ice, dry ice-ethanol, and liquid nitrogen traps. Since this technique was adopted as the method of flavor volatile recovery from the HTST fluid sterile milk to be analyzed in this research program, it will be discussed more specifically in the procedure section. As evidenced by this short review of the more note- worthy volatile collection techniques, numerous and imagina— tive system designs have been employed in flavor volatile extractions. Although each has attempted to eliminate the artifacts which may involuntarily arise during analysis, no one system to date has been developed which eliminates all of the shortcomings found in our present recovery systems. Means of Identification of Isolated Components Gas liquid chromatography provides a suitable means of separation and tentative identification of volatile flavor components by use of retention volume data. To the investi- gator who is working with an entirely unknown mixture of compounds, these data have limitations, since some components of the mixture, upon gas chromatographic separation, may exhibit the same, or nearly the same retention data, and yet possess an entirely different functional group (Walsh and Merritt, 1964). This problem has been resolved by collecting the gas chromatographic effluents or, as is more recently the case, splitting the sample stream.to the gas chromatographic detector, and employing infra-red spectro— metry (Billis and Slowinski, 1956; Wheaton and Wentworth, 1959; Chang et a1., 1961) mass spectroscopy, or time-of-flight mass spectrometry (Ebert, 1961; Gohlke, 1959; Linderman and Annis, 1960; McFadden et al., 1963), for more conclusive identification. Because of the cost of such instrumentation, this type of confirmatory analysis is beyond the reach of most researchers and identification of volatile flavor com- ponents must be made by gas chromatographic retention data, coupled with functional group analysis. Thin—layer and paper chromatography have also proved to be invaluable tools in . this work. Functional group analysis is basically a rather un- complicated technique which consists of reacting aliquots of an unknown chemical mixture with Special reagents to confirm or negate the possible presence of aldehydes, ketones, alcohols, esters, mercaptans, amines, and sulfides. Alde- hydes can be detected by Schiff's reagent, 2,4 dinitrophenyl- hydrazine (Bassette et al., 1962; Walsh and Merritt, 1960: Lynn et al., 1956; Badings and Wassink, 1963), ketones by 2,4 dinitrophenylhydrazine, and acid hydroxyl amine (Bassette et al., 1960; and Badings and Wassink, 1963), alcohols by 3,5 dinitrobenzoates, ceric acid (Cheronis and Entrikin, 1957; and Heff and Feit, 1964), esters by ferric hydroxymate, basic hydroxyl amine, and sodium hydroxide (Walsh and Merritt, 1960; Bassette et al., 1962; Hoff and Feit, 1964), mercaptans by alcoholic silver nitrate, lead acetate, isatin, and sodium nitroprusside (Walsh and Merritt, 1960; and Shriner et al., 1956), and sulfide with mercuric chloride (Bassette et al., 1962). Thin-layer chromatography has also proved to be an 10 invaluable aid in the identification of the carbonyl components of flavor isolates as shown by Parks et a1. (1964), and Badings and Wassink (1963) who have established an extremely refined procedure for the separation of the homologous classes of both aldehydes and ketones. This technique coupled with the method for class separation by absorption-chromatography developed by Schwartz et a1. (1962), and Schwartz et a1. (1963), has endowed the flavor chemist with a reliable tool for identification of these flavor components. More recently Soukup et a1. (1964) accomplished the identification of 2,4 dinitrophenylhydrazones by gas chromatographic analyses. This latter technique is especially useful since derivatives need not be purified and trace amounts of these compounds can be detected. Recently Merritt and Walsh (1962) developed a new method of functional group classification and subsequent qualitative identification of chromatographic peaks by the use of dual column gas chromatography in which each of the two columns has a different liquid phase column packing. This enables qualitative identification of the peak to be made solely by means of retention volume constants. Flavor Components of Milk and Milk Products In the past decade researchers in the field of dairy chemistry have enjoyed much success in isolating and deter- mining natural and off-flavor components in dairy products. The reason for these many advances has been the development of new instrumental techniques. 11 The majority of the compounds isolated and identified have been carbonyl in nature, and many have been positively linked to certain natural desirable or undesirable flavors. The following is a chronological summary of the known com- pounds that have been isolated, lated these compounds, and where known, flavor with which each has been associated. the worker(s) who have iso- the characteristic Table l. Chronological review of flavor components isolated from stored dairy products. Com ound Source Flavor Worker(s) and p Defect Reference No. 2 hydroxypropanal Oxidized oxidized Keeney and Doan C7 and C9 unsaturated butteroil (1951a, 1951b, ketones 1951c) a compound with the empirical formula C12 H20-22O C7-C9n-alkanals dry whole Shipstead and milk (old Tarassuk (1953) sample) C4—Cll a1k~2—enals oxidized oxidized Forss et a1. C6-C9 alk 2,4 dienals skimmilk (1955a, 1955b) acetone ethanal n—hexanal crotonaldehyde methyl sulfide fluid milk cowy Patton, Day, flavor and Forss (1956) delta~deca1actone evaporated stale and Keeney and delta-dodecalactone and dry coconut Patton (1956a, whole milk 1956b) ethanal gamma Day et a1 (1957) acetone irradiated butanone skimmilk formaldehyde n-hexanal methyl alcohol 12 acetone all odd number C5-C 5n—a1k—2-ones but - 2-ena1 Table 1. Continued. Flavor Worker(s) and compound Source Defect Reference No. ethyl alcohol methyl mercaptan methyl sulfide furfural heated stale Morgan et a1. acetaldehyde skimmilk (1957) acetaldehyde evaporated typical Dutra et a1. 2-pentanone milk evaporated (1959) 2—heptanone flavor 2-pentanone evaporated Wong et a1. 2-heptanone milk (1958) acetone C3-C15 methyl ketones heated heated Patton and unheated milkfat— milk fat Tharp (1959) acetone only formaldehyde acetaldehyde acetone butanone methylpropanal nonfat storage Bassette and 3-methy1butana1 dry milk stale Keeney (1960) furfural diacetyl hexanal nonanal . C5-C10n—alkanals washed oxidized Forss et a1. C5-C10alk 2—enals cream (1960b) 2,4-heptadiena1 2-heptanone C1-ClOn-a1kanals autoxidized C5-Cllalk-2-enals milkfat oxidized Day and Lillard (1960) Table 1. Continued. 13 Compound Flavor Source Defect Worker(s) and Reference No. n-hexanal n-heptanal hex-2-enal 2-heptanone 2-pentanone 2-nonanone 2-undecanone acetone C C , C -C alkanals 3 5 8 10 alk-2-enals CI7IC'C C§7C9alk-§-ongs hepta-2,4 dienal acetone Cl'ClOn-alkanals C5-Cllalk-enals ethanal methyl sulfide acetone C -C3( Cs'ClO’ Clzn a kanals C3-09: C11: 013: C15 methyl ketones C5—C11 alk—2-enals cis furfural trans furfural benzaldehyde ethanal acetone 2—pentanone a-pentyl acetate dimethyl sulfide oct-l—en-3-one ethanal acetone butanone 3—methyl butanal 2-methyl propanal milk fat fishy flavor milk fat fluid milk dry whole milk (badly deteriorated powder) sterile concentrated milk butteroil metallic normal flavor raw milk Forss et a1. (1960) Day and Lillard (1960) Wynn et al. (1961) Parks and Patton (1961) Patel et a1. (1962) Forss, et a1. (1962) Wood and Aurand (1963) Table 1. Continued. l4 Flavor WOrker(s) and compound Source Defect Reference No. alk-2—enals milk fat oxidized Keith and Day alkanals (1963) alk-2,4—dienals C5-Cl6n-a1kanals butteroil Parks et a1. C6-Cllalk-2-enals from (1963) C3-C12alk-2,4 dienals spontaneously oxidized whole milk formaldehyde stale dry stale Nawar et a1. propanal whole milk (1963) ethanal acetone 2—butanone or 2—pentanone a compound similar to heptaldehyde an unsaturated dicar- bonyl or hydroxycarbonyl methyl sulfide cultured Day and Lindsay cream or (1963) butter 2-pentanone evaporated storage Muck et al. 2-heptanone milk 2-nonanone 2-undecanone caprylic acid 2 tridecanone delta-decalactone delta-dodecalactone gamma-dodecalactone methyl sulfide fresh natural Wong (1963) acetone cream butanone ethanol isovaleraldehyde ethyl alcohol chloroform 3-methoxy—4-hydroxy— heated carmel Cobb et a1. benzaldehyde milk flavor (1963) 15 Table 1. Continued. C d S r Flavor Worker(s) and ompoun ou ce Defect Reference No. formaldehyde fresh normal Winter et a1. ethanal butter flavor (1963) isobutyraldehyde isovaleraldehyde n-hexanal n-nonanal phenylacetaldehyde acetone 2—heptanone 2—nonanone diacetyl acetoin C3-C8 alkanones sunlight Wishner and Cl-C 2 alkanals exposed Keeney (1963) C , C -C alk-2—enals milk 4 12 dimethyl sulfide butter normal Day et a1. (1964) oct-l-en—3-ol oxidized mushroom Starkand Forss milk and (1964) butter n-hexanal milk fat fishy Forss (1964) n-heptanal flavor 2-hexanal 2-heptanone n-amyl vinyl ketone oxidized metallic Forss (1964) fat flavor ethanal sterile Bingham (1964) propanal acetone butanone 2-pentanone methyl sulfide milk Toan et a1. acetaldehyde acetone (1965) 16 Table 1. Continued. Compound Source Flavor Worker(s) and Defect Reference No. 2-heptanone 2—nonanone Zdundecanone 2-tridecanone sterile stale Arnold et a1. benzaldehyde concen- (1966) napthalene trated alpha-dichloro- milk benzene benzothiazole ortho-aminoacetophenone delta-decalactone A recent review by Day (1965) also carries a complete list of the carbonyl compounds from flavor isolate of dairy products along with the melting points and ultraviolet ab- sorption maxima of their 2,4 dinitrophenylhydrazone deriva- tives. Origin of the Flavor Compounds In dealing with milk that has been heated to a temperature of 750C. or above, even momentarily, the first unnatural off-flavor that one encounters is the so-called cooked flavor as described by Harland and Ashworth (1945). 17 The intensity of this cooked flavor, as shown by Harland and Ashworth (1945) is a function of two parameters: degree of heating and duration of the heating interval. This flavor arises from compounds liberated by the heat de- naturation of the milk serum proteins as shown by Harland and Ashworth (1945), Harland et a1. (1955), Hutton and Patton (1952), Meneffe et a1. (1941), Rowland (1937), and Rowland (1938). The particular protein involved has been shown by Larson and Jenness (1950) to be the B-lactoglobulin fraction of raw milk serum proteins. Hutton and Patton (1952) later confirmed B-lactoglobulin as the primary source of sulfhydryl groups while the fat globule membrane material was also shown to be a minor contributor. Concomitant with the appearance of the reactive sulfhydryl groups in heated milk there also appears a re- duction in the oxidation-reduction potential of the milk. The inhibitory effect of heat treatment of milk on oxidative deterioration has been reported by the following workers; Hutton and Patton (1952), Gould and Sommer (1939), Josephson and Doan (1939), Gould and Keeney (1957), and Patton (1955), and has been attributed to the preferential oxidation of the activated sulfhydryl group rather than the unsaturated fatty acids. The exact mechanism of sulfhydryl oxidation and the reaction products of sulfhydryl oxidation are at this time still unknown. The origin of the sulfides and mercaptans which are present in milk is also not completely understood, but according 18 to Patton (1955) and Dutra et a1. (1959) sulfhydryl groups can be converted to volatile sulfides, mainly hydrogen sulfide. The presence of methyl sulfide in milk has been attributed by Patton et a1. (1956) to methyl sulfone which is present either as a normal constituent of cow's blood or by bacterial decomposition of algae in the rumen of the cow. Recently Toan et a1. (1965) have shown that a high Aerobacter aerogenes population can also cause an increase in the methyl sulfide content of milk. Day et a1. (1964) have suggested that dimethyl sulfide can be produced from three precursors: a) normal metabolic breakdown of methionine, which has been proven to be the production route from normal dimethyl sulfide levels. b) dimethyl-beta propiothetin, c) and methylmethionine- sulphonium salt. The two latter precursors are believed to account for excessive dimethyl sulfide levels. These pre- cursors are of plant origin and would have to be transmitted via the rumen to the milk. This pathway for transmission of such volatiles has been demonstrated by Dougherty et a1. (1962). Lipid deterioration in milk can be caused by lipolysis and/or autoxidation, both of which can cause development of off-flavors. In this work, however, our concern will be with oxidative deterioration of the lipid phase and the secondary products of this oxidative mechanism, aliphatic aldehydes and ketones both saturated and unsaturated, and alcohols. Saturated and unsaturated aldehydes have been shown to impart characteristic off-flavors to milk at very low concentrations 19 (Boldingh and Taylor, 1962; Wong and Patton, 1962; Patton and Thorp, 1959; Van der Ven et al., 1963: Wishner and Keeney, 1963; Parks et al., 1963: Day et al., 1957; Forss, 1964: Forss et al., 1960). Day et al., (1963) have presented evidence concerning the additive interaction exhibited by these carbonyl compounds to give a detectable flavor even though all of the carbonyls were present at subthreshold concentrations. The origin of these aldehydes has been the subject of research for many years, and one of the most popular and substantiated theories for aldehyde formation is that of autoxidation of the lipid fraction of milk (Stull, 1953; Badings, 1960; and Schultz et al., 1962). Although little is known concerning the initiation step in the autoxidation of the unsaturated fatty acids to form free radicals, Farmer and Sutton (1943) have elucidated the resulting propagation reactions, which yield hydroperoxides, and free radicals capable of propagating the reaction. In addition to hydro— peroxides, peroxides are also known to arise from this mechanism (Parks et al., 1963: Lewis and Quackenbush, 1949: and Willites, et al., 1953). Hydroperoxides in themselves are bland and odorless compounds causing no off—flavor; however, these hydroperoxides are extremely unstable compounds and their dismutation leads to the formation of carbonyls, alcohols, and epoxides, as secondary reaction products. A good review of this topic has been written by Badings (1960). 20 Winter et al. (1963) postulated that some aldehydes present in butter may originate from amino acids by enzyme- catalyzed oxidative decarboxylation. If this hypothesis is correct, then formaldehyde, acetaldehyde, isobutyraldehyde, isovaleraldehyde, phenylaldehyde might originate from glycine, alanine, valine, leucine, and phenylalanine respectively, which occur in the proteins of milk. Depending upon the previous history of the milk some of these amino acids may also exist as free amino acids. Dutra et a1. (1959) in support of this theory have shown that acetaldehyde can in fact originate from alanine. Wong (1963) has demonstrated that isovaleraldehyde, which gives rise to a malty flavor, may arise from the metabolic breakdown of leucine by s, lactis var. maltigenes. Many theories exist as to the identity or the contri— bution of various classes of aldehydes in causing oxidized, and stale—oxidized flavor in processed dairy products. Parks et a1. (1963), for example, in studying carbonyl com- pounds in butteroil from nonoxidized and spontaneously oxidized milk, indicated that alk 2,4 dienals (especially 2,4 decadienal) play an important role in off-flavor development, whereas the saturated aldehydes have slight, if any, significance in this role. Forss et a1. (1960) and Bassette and Keeney (1960) on the other hand, have shown the presence and flavor signifi- cance of the lower saturated aldehydes in oxidized dairy products. These findings have been collaborated with the research of Nawar et a1. (1962), who identified the C to C 1 9 21 saturated alkanals in oxidized milk fat, and postulated their contribution, using the additive interaction theory of Day et a1. (1963), as a basis for their suppositions. There is uncertainty as to the part played by the C1 to C12 saturat- ed aldehydes in oxidized flavor. All workers seem to be in general agreement concerning the involvement of the mono- and di-unsaturated alkenals along with the branched chain aldehydes in the oxidized flavor of dairy products, as shown by the research of Bassette and Keeney (1960), Day and Lillard (1960), Parks et a1. (1961), and Forss (1955a, 1955b). Origins of the saturated, unsaturated, and branched chain aldehydes have been determined with a fair degree of certainty. Kawahara (1952), Farmer and Sutton (1943), Badings (1960), Day and Lillard (1960) and Gaddis et a1. (1961) have been unable to show oxidative reaction schemes for C C C C C C C alkanals; C C C C C 2' 3' 4' 6' 8' 9' ll 4' 5' 6' 7' 8' 2,5 C11, 3,6 C12 alkdlenals; 2,4,7 C10, 2,4,7 C13, 2,5,8 Cl4 alktrienals; and finally 2,4,7,10 C alketetraenal, from 16 oleic, linoleic, linolenic, and arachidonic acid. Parks et a1. (1963) have shown that the Cll‘cls saturated aldehydes present in milk could arise from two possible sources: a) synthesis of milk fat and/or b) hydrolysis of plasmalogens or neutral plasmalogens during the process of pasteurization. Investigations by Van Duin (1958) and Parks et a1. (1961) seem to support this theory by showing that the Cll to C18 saturated aldehydes are bound to glycerol of the plasmalogen fraction in butteroil. 22 The branch chained aldehydes that have been found to exist in milk systems have not been identified with any type of lipid degradation. The suggestion has been made that they are products of the Strecker degradation of their respective amino acids (Bassette and Keeney, 1960). Jackson and Morgan (1954) have shown conclusive evidence for 3 methylbutanal formation by Strecker degradation of leucine by g, lactis var. multigenes. Dutra et al. (1958), using radioactive tracers, have elucidated two additional pathways for acetaldehyde formation: a) lactose breakdown, and b) deamination-decarboxylation of alanine. The other group of carbonyl compounds that is respons- ible for off-flavor development in milk and milk products is the ketones. Ketones with both saturated and unsaturated structures have been shown to impart characteristic off- flavors, even at very dilute concentrations (Bingham, 1964; Wong et al., 1958; Nawar et al., 1962: Wong and Patton, 1962; Patton and Tharp, 1959; Langler and Day, 1964: Parks and Patton, 1961; and Schultz et a1. 1962). Langler and Day (1964) also pointed out that ketone mixtures exhibited the same synergistic interaction as was shown for aldehydes. Methyl ketones, one class of the volatile ketonic flavor compounds that have been isolated from pasteurized. concentrated, and sterile concentrated milks (Bingham, 1964: Boldingh and Taylor, 1962; Wong et al., 1958; Muck et al., 1963; Wong, 1963: Dutra et al., 1959; and Winter et al., 1963) 23 have very controversial origins. The fact that methyl ketones have been isolated from milks that have had no heat treatment seems to give evidence that their formation is easily initiated. Acetone is the only methyl ketone that has been consistently isolated from unheated dairy products, and it is thought to be present from the time of secretion as shown by Patton and Tharp (1959). Boldingh and Taylor (1962) postulated that the methyl ketones have as their origin the beta-keto acids bound to the glycerides. Wong and Patton (1962) presented evidence to support the Boldingh-Taylor theory, when they showed that methyl ketones could be easily produced by spontaneous de- carboxylation of beta-keto acids, which are intermediates in the beta oxidation of fatty acids. Recently, van der Ven et a1. (1963) have presented evidence in support of this theory by demonstrating that there are beta-keto esters in milk and that methyl ketones are formed in the presence of heat and moisture. He suggests that the C5-C15 methyl ketones are derived from the C6-Cl6 beta-keto acid glycerides. This work was accomplished by reacting the milk fat with Girard-T- reagent and subsequently identifying the pyrazoles. These data have been substantiated by Parks et a1. (1964), who have isolated and identified beta—keto esters as the precursors of methyl ketones in milk fat. It has also been shown by Wong et a1. (1958), Patton and Tharp (1959), Parks et a1. (1964), Langler and Day (1964), and Schwartz et a1. (1965), that prior to formation of methyl ketones in milk fat, hydrolysis must be effected whether by heating in the presence 24 of water or by saponification. Evidence presented by Patton and Tharp (1959), showing that fresh unheated milk fat is devoid of such ketones, and the latter findings of Langler and Day (1964), concerning the inhibition of methyl ketone formation when milk fat was dried over calcium hydride for 18 hours, also seems to lend support to this hydrolysis theory. Nawar et al. (1962), have studied heat—induced changes in milk fat and did not show the necessity for the presence of moisture in formation of either even or odd numbered methyl ketones in freshly prepared butteroil: however, no mention was made of completely drying the oil before heating was initiated, so that moisture present in the oil could have been sufficient for hydrolysis. These re— searchers also feel that methyl ketone formation could not occur as a result of atmOSpheric oxygen attacking the lipid phase because of the abundance of reducing substances present in commercially processed dairy products. a,8 unsaturated ketones have been shown by Ellis (1950) to arise as the first reaction product in metal catalyzed lipid oxidation rather than the d,8 unsaturated hydroperoxide proposed by Farmer et a1. (1943). Ellis also postulated that these unsaturated ketones complex with molecular oxygen to form a compound having properties of peroxides. The traditionally accepted theory is that the a,8 unsaturated ketones arise from the dismutation of the hydroperoxides formed by the mechanism described by Farmer et a1. (1942). 25 Staling, a deteriorative reaction which takes place in dry and fluid milk products, has never fully been defined. Supplee (1926) as long ago as 1923, recognized the stale, musty, or gluey flavor in stored, high—moisture milk powders. Lea et a1. (1943) have termed the stale flavor that developed in dry whole milk as heated, burnt, scorched, or cooked, and in 1962 Patel et a1. (1962) described this condition as "old- rubber" in sterile concentrated milk. Lea et a1. (1943) believed that stale flavor consisted of two compounds: a) the burnt, or carmel taste, which they associated with the protein or carbohydrates; and b) a "butter- toffee” flavor which they felt was associated with the fat phase of dry whole milk powders. Whitney and Tracy (1949, 1950) fractionated stale dry whole milk into cream, skim- milk, butter, buttermilk, butteroil and butter serum, and concluded that stale flavor is concentrated in the fat phase. They were careful to point out that this was not evidence as to the origin of the flavor. Whitney and Tracy (1949) also determined that there was a greater concentration of stale flavor per unit weight in the products that were high in phOSpholipids. Bassette and Keeney (1960), in studying the storage defect associated with non-fat dry milk, indicated that both Maillard browning and lipid oxidation were involved in this flavor deterioration. Evidence was presented for the presence of both odd and even chained aldehydes and ketones, and for the presence of branched chain aldehydes. Furfural and diacetyl 26 were also identified as reaction products of the Maillard browning reaction. These data also agree with the later research findings of Nawar et a1. (1963) who felt that there were at best two different components capable of inducing the stale-flavor sensation, one component was extractable as a Girard-T-derivative and was shown by infrared absorption spectroscopy to contain two functional groups: a) one in the lactone-ester region, and b) one in the ketone—aldehyde region. The second flavor component capable of eliciting a stale- flavor response was found in the carbon tetrachloride vapor from the volatile extraction technique and appeared to contain both saturated and unsaturated carbonyl compounds. The data from these latter two flavor fractions agree with the findings of Parks and Patton (1961) who suggested that staleness may be caused bya combination of compounds, being either aldehydes or ketones, and their concentrations determining which flavor was elicited; stale or oxidized, the former being caused by a lower carbonyl concentration than that necessary for oxidized flavor. The origin of these carbonyls has been shown earlier as secondary reaction products of the oxidized triglycerides and phospholipids in milk fat. The presence of the lactone—ester region in the infra- red spectrophotometric analysis of stale flavor isolate has given support to the postulation that the coconut and fruity flavors exhibited in early storage of dairy products originate from lactone formation. Keeney and Patton (1956a, 1956b) identified the origin of the coconut flavor as delta- decalactone. Tharp and Patton (1960) identified the compound 27 contributing to the fruity flavor to be delta-dodecalactone. In both cases these flavor compounds are believed to arise from action of heat on the milk fat, since no lactones were found in unheated milk. This theory was later affirmed when Mattick et a1. (1959) showed the reaction scheme for delta-decalactone formation from 5—hydroxydecanoic acid. which exists in native butterfat as a simple ester. This same research also demonstrated that both the degree of heat treatment in the processing technique and temperature of storage are of primary importance in the development of the delta-decalactone from 5—hydroxydecanoic acid. Muck et al. (1963) and Arnold et a1. (1966) have also showed the involvement of lactones in stale flavor development by isolating them from stale evaporated milks. The last cause of off-flavor development to be con- sidered is that of non-enzymatic, or Maillard browning, which develops during the storage of dairy products. Tarassuk and Jack (1948) also pointed out that the naturally occurring phenomenon of non-enzymatic browning could be related to the development of stale and oxidized flavors in processed milk products. The mechanism postulated for this browning reaction by the French chemist Maillard, and discussed by Lea (1950) is concerned with the interaction of carbohydrates with amines and free amino acids. Harland et a1. (1947), Kass and Palmer (1940), Patton (1952), and Patton and Josephson (1949a, 1949b) showed that in milk systems the reactants were lactose and 28 casein. Evidence from Patton (1952), and Patton and Flipse (1953) pointed to the epsilon-amino group of lysine as being the key reactant group of casein in the browning mechanism. The mechanism of browning in the sugar—amine systems is extremely complex. The literature concerning these reactions in model systems reveals seven different types of reactions that have been shown to take place; however, the extent of which each is present in the natural browning of products is not known. The initial stage of the reaction involves sugar- amine interaction yielding an N—substituted glycosylamine. It is postulated by Lea and Hannan (1950) that these N- substituted glycosylamines isomerize to l-amino-l-deoxy-Z- ketoses, which is known as the Amadori rearrangement. The intermediate stage in the sugar—amine browning systems follows the Amadori rearrangement and as discussed by Hodge (1953) can lead to sugar dehydration, sugar fragmentation, formation of furfurals, reductone-like reducing substance, fluorescent substance, and the degradation of amino acids by the Strecker reaction. The final stage of browning is a polymerization reaction to form unsaturated, fluorescent, colored polymers. The main reactions of this polymerization reaction as dis- cussed by Hodge (1953) are thought to be aldol condensations, aldehyde—amine polymerization, and the formation of hetero- cyclic nitrogen compounds, called melanoidins. Hodge (1953), Hodge and Rist (1953) along with Lea and Hannan (1950a, 1950b) have discussed at length these reactions and how they fit into the findings concerning the casein-lactose complexes. 29 The most important manifestation of the non-enzymatic browning system is the formation of the browning pigments, melanoidins, which according to Kass and Palmer (1940) and Patton (1952) are chemically bound by milk proteins and are responsible for the discoloration and flavor of milk systems. Patton (1955) has also proposed that when milk reaches this stage of discoloration there are at least four flavor compo— nents associated with it: 1) carmel or malty, resulting from sugar decomposition and Strecker degradation of amino acids, 2) "stewed meat", arising from methionine decomposition, and the presence of hydrogen sulfide which arises from the presence of reactive sulfhydryls, 3) hydrolytic rancidity, arising from fat hydrolysis, and 4) coconut, resulting from the formation of lactones in the milk fat phase as previously discussed. More detailed discussions concerning the origins of these flavor compounds can be found in the publication of Hodge (1953), Hodge and Rist (1953) and Patton (1955). EXPERIMENTAL PROCEDURE The sterile milk used in this research project was produced by a commercial process using the DeLaval Vacu-Therm Instant Sterilizer, which is a high—temperature, short-time steam injection sterilization unit. In this unit the raw milk was preheated to 770C. and then instantaneously heated by steam injection to 140.50C i 10C. Following an approximately 4.0 second hold at this temperature the sterile product was then cooled to 77°C. in a reduced—pressure vessel. The desired post-sterilization temperature was achieved by controlling the vacuum on this vessel by a hand operated release valve. The purpose of this vacuum cooling procedure was to remove the water added by the steam injection process and thus assure a proper total solids/ water ratio in the sterile product. At this point the milk was homogenized in an aseptic Manton Gaulin homogenizer at a pressure of 4000 psig. (3000 psig. on the first stage and 1000 psig. on the second stage). Proper control of temperature and pressure are essential to assure satisfactory homogeniza- tion, and minimizes the development of cream layer and sediment in the sterile product during storage. Following homogenization the milk was cooled to a temperature of 12.5°C. in an aseptic plate heat exchanger and then piped directly to an aseptic balance tank to await sampl- ing, or further processing. 30 31 The procedure for sterilization of the DeLaval VTIS system was modified from that programmed in the automatic sterilizing cycle. This modification resulted from a lack of steam pressure on the final aseptic balance tank, which was observed to be 7 psig. Since the sterilizing program was based on an F of 15 minutes, it was necessary to calculate 0 a new F234 with a z of 180F., which resulted in a sterilizing period of 3.9 hours. Sample Collection Techniques One gallon glass bottles were used as the storage containers for the sterile fluid milk samples. The procedure for cleaning and sterilizing these bottles was as follows: the bottles were washed twice in detergent solution, thoroughly rinsed in tap water, rerinsed with distilled water, and loosely capped. The bottles were then steam sterilized for 20 minutes at 15 psig., after which the bottles were cooled and the caps securely sealed. Thirty bottles were used for each set of samples taken, and there were three sets of samples collected. Thirty presterilized bottles were then placed in a 48" x 24" x 30" Plexiglass dry-box for further processing. This dry-box was constructed with a gas inlet at each end, two arm holes fitted with rubber gloves in the front panel. a one-half inch hole in the back panel through which a sterile sampling hose entered the chamber and a sterile air filter fitted to the top of the box. The purpose of this dry-box 32 was to provide a sterile atmosphere in which sampling could be carried out and thus reduce the possibility of air—borne contamination. Sterilization of the atmOSphere in the chamber was accomplished with the use of the sterilizing gas, ethylene oxide, by the following procedure: a hose leading from the pressure reducing valve of the ethylene oxide-carbon dioxide gas cylinder was connected to one of the gas inlets, both inlets were then opened, and the pressure on the reducing valve adjusted to approximately 3 psig. The system was permitted to remain under these conditions for five minutes to assure complete purging of the atmosphere in the dry-box. At various intervals during this purging operation the gas jet used as an exhaust vent was closed for 15 to 20 seconds and then reopened, thereby accelerating the exchange of atmospheres in the sampling chamber. At the completion of the five minute purge interval the inlet of the sterile air filter was capped, the gas outlet closed, and the pressure of the ethylene oxide-carbon dioxide gas mixture in the chamber adjusted to 3 psig. At this point the gas inlet was closed and the chamber and its contents underwent sterilization for eight hours. Ethylene oxide is both explosive and toxic, and, since a concentration of 50,000 ppm. is not tolerable for more than a few minutes, a well- ventilated room should be used for such research. Upon completion of the sterilization period a vacuum pump was connected to one of the gas jets. The jet was 33 opened, the pump started, and the cap was then immediately removed from the inlet of the sterile air filter. The system was evacuated for two hours in order to replace the ethylene oxide-carbon dioxide gas in the sterile chamber with sterile air. This was done to preclude the presence of ethylene oxide in the headspace gas of the samples which could react with the milk fat during storage. Completion of this evacuation phase completed the sterilization procedure, and with the attachment of a presterilized hose to the sampling valve of the aseptic balance tank, the system was ready for sampling. During sampling the sterile bottles were uncapped a few seconds before they were to be filled and were immediately recapped after filling. After all of the samples had been taken the sampling valve was closed, the hose disconnected and the bottles re- moved from the sampling chamber. They were then placed in storage cabinets maintained at 4.4°C., 22.00C., and 360C. to await analysis at predetermined intervals of zero, one, two and three months. Volatile Recovery The recovery of flavor components from fluid sterile milk was accomplished with a low-temperature, reduced-pressure distillation technique described by Day et al. (1957) for recovering the volatiles from gamma-irradiated skimmilk. The standardized procedure for recovery of flavor components from both fresh and storage fluid sterile milk 34 was as follows: six liters of the sterile product at a temperature of 100C. were placed in a 12-liter distillation flask. 186 mg. of Dow—Corning Antifoam AF Emulsion was added and the pressure in the system was reduced to 20 to 30 mm. mercury. Nitrogen gas was used to control the pressure and afforded three benefits: a) the bubbling of the nitrogen into the system provided good agitation, b) a nitrogen atmOSphere prevented oxidation during distillation, and c) the addition of nitrogen acted as a sweep gas to carry off both the water vapor and flavor volatiles from the distillation flask. No heat was applied to the distillation system until the operating pressure of 20 to 30 mm. mercury was reached, at which time the temperature was immediately raised to 400C. 1 10C. The sample was then allowed to distill for a period of four hours and the distillate from the sterile milk was fractionally collected in various traps: wet ice, dry ice and ethanol and liquid nitrogen. Upon completion of the distillation period the distillation flask was removed from the system and the volatiles from the traps were distilled over to the final liquid nitrogen trap. This was accomplished by salting out each distillate trap with anhydrous sodium sulfate and then varying the temperature and pressure on the trap until the volatiles were moved from the aqueous distillate to the next trap. This procedure of transferring all volatiles to the final liquid nitrogen trap was deemed standard procedure after previous gas chromatographic analyses showed that no volatiles 35 were inijmawet ice or dry ice traps that were not present in the liquid nitrogen trap. Following the distillation and concentration techniques in the volatile recovery procedure the volatiles were kept in liquid nitrogen until gas chromatographic analyses were performed. Gas Chromatography Gas chromatographic analyses were performed on an F & M model 810 research gas chromatograph equipped with a dual flame ionization detection system, a model 50 automatic attenuator, and a column oven designed for automatic temperature pro— gramming from ambient to 400°C. The columns used were prepared in a manner outlined by Dal Nogare and Juvet (1962). A six foot length of 1/4 inch 0. D. copper tubing was packed with 25% w/w carbowax 20 M on acid—base washed 80/100 mesh Chromosorb W. Packing was accomplished with the aid of a Vibra—graver tool and both columns were closely matched to insure minimal baseline drift during the temperature programming sequence. Following the packing of both columns they were installed in the model 810 chromatograph and conditioned for one week by continually programming the oven temperature from ambient to 210°C. at a rate of 1°C. per minute for two days, and then holding the temperature at the upper limit for the remaining five days. At this time the helium flow rate in both columns was set at 50 ml/min. The instrument was then checked for drift during 36 a programmed run of 4°C/min., and the helium flow on the reference column, B column, adjusted so as to balance the background signal emitted by the analytical column at the upper limit of the temperature program. At this point the instrument was used for analysis of the flavor distillates. The following conditions were established for the gas-liquid chromatographic analysis of the extracted flavor distillates: range setting ----------------- lx attenuation ------------------- lower limit 4 recorder sensitivity ---------- l mv. temperature ------------------- 12 minutes isothermally at 70°C., then 4°C. per minute to an upper limit of 210°C. detector temperature ---------- 285°C. injection port temperature----250°C. detector. ---------------------- hydrogen flame flow rates: carrier gas -------- helium 50 ml/min. hydrogen ----------- 63 ml/min. compressed air ----- 500 ml/min. recorder speed ---------------- 0.25 inch/min. column data ------------------- length 6 feet 1/4 inch 0. D. 25% carbowax 20 M on acid-base washed 80/100 mesh Chromosorb W. Sampling of the Flavor Isolate for Analysis The glass stopper was removed from the cold finger, while it was still submerged in liquid nitrogen, and a rubber stopper, fitted with a rubber septum and covered with a double layer of parafilm, was inserted in its place. The cold finger was then removed from the liquid nitrogen and warmed to a temperature of .70°C. It was important to warm the cold trap slowly and remove some of the pressure from the trap by periodically inserting a hypodermic needle through the septum. 37 Preliminary analyses demonstrated that no volatiles were lost through this technique. After the temperature of the trap reached equilibrium with the temperature of the water bath a gas tight syringe was used to remove a 6 ml. gas aliquot from the trap. The sample of gas was then injected into the chromato- graph. Relative retention times were obtained for the isolated components from the flavor distillate, and these were always run in duplicate. Often the individual flavor constituents are similar, or are structural isomers, or have the same relative retention time as another compound. Thus, in order to make some tentative identification as to the components present, it was necessary to perform functional group analyses. The technique of Hoff and Feit (1964) in which these reactions were carried out in the vapor sampling syringe was adapted because of its ease and efficiency. The only modification that was found necessary was the use of 1 ml. of reagent for proper reaction. Reactions for aldehyde, ketone, ester, alcohol, and various sulfur compounds were performed on the distillate vapor from each extraction. These chromatograms were then compared with those of the untreated vapor to aid in establishing the functional groups and the identity of the compounds present. Once the information had been collected from these chromatograms these data were compared to the relative retention times of known compounds that had been chromato- graphed under similar conditions. Using all of this information an effort was made to tentatively identify some of the components 38 that were responsible for the stale—oxidized flavor that developed during storage. Thin Layer Chromatography Another technique that was utilized to determine the identity of these off-flavor compounds was thin-layer chromato- graphic identification of the 2,4 dinitrophenylhydrazone derivatives of the carbonyls present in the flavor isolate. The hydrazone derivatives were formed using a modified procedure of Lawrence (1965), in which a 2:1 ratio of 2,4 dinitrophenyl- hydrazine reagent to aqueous flavor distillate emu; mixed in a glass stoppered bottle and allowed to remain quiescent for 12 hours. The 2,4 dinitrophenylhydrazones were then extracted from the mixture with 30 ml. of carbonyl-free hexane. prepared by the method of Schwartz and Parks (1961). The aqueous layer and upper hexane layer, containing the 2,4 DNP derivatives, were then permitted to separate in a separa- tory funnel. The hexane extraction was then repeated using 20 m1. aliquots of carbonyl—free hexane until no color was apparent in the hexane. All hexane extractions were then collected and evaporated to dryness using a rotary evaporator. Because of the small quantities of 2,4 DNP hydrazones present, meltingCX'mixed melting points could not be done. Identifi- cation of the 2,4 DNP hydrazones was then made by thin layer chromatography of the homologous classes of carbonyl derivatives. The class separation performed on the mixture of 2,4 DNP hydrazones was that described by Schwartz et a1. (1962) using 39 a Magnesia-Celite 545 column. This gives the elution sequence of methyl ketones, saturated aldehydes, 2-enals, and 2,4— dienals. These various classes, which could easily be identified on the column because of the colored bands they exhibited, were then analyzed spectophotometrically to deter— mine their absorption maximum, which assists in identification of the eluting class. These homologous classes were then separated and identified using the thin—layer chromatographic techniques reported by Badings and\Wassink (1963) for the separation of 2,4 DNP hydrazones on Kieselguhr G-Carbowax 400 impregnated plates. All known 2,4 DNP hydrazone derivatives of the carbonyls used for identification of unknown were prepared and doubly recrystallized according to the procedure discussed by Shriner and Fuson (1948). Bacterial and Organoleptic Analyses of Storage Sterile Milk Samples All HTST fluid sterile milk samples, prior to the recovery of the flavor components by low-temperature, reduced- pressure distillation were analyzed for the presence of bacteria by the Standard Plate Count Method. Standard plate counts were made utilizing the technique described in Laboratory Manual-Methods of Analysis of Hulk and Its Products. A one milliliter sample of undiluted milk was plated since these analyses were made merely to ascertain whether or not the milk was sterile at the time of flavor extraction, and thus assure that none of the flavor compounds isolated were caused 40 by bacterial decomposition of the milk during storage. The HTST fluid sterile milk was also subjected to organoleptic evaluation prior to the flavor distillation of the volatiles. The fluid sterile milk was graded on a numerical basis using the grading system set forth by the American Dairy Science Association and the Dairy Industry Science Association for scoring milk and milk products. The normal range in grading market milk according to this system is from 31 to 40 points with 31 showing a product of near unsalable quality and 40 denoting an excellent product devoid of criticism. A breakdown of the scoring range for the broad classification of excellent, good, fair, and poor is as follows: excellent, 40; good,38-39.5: fair, 36-37.57 poor, 35.5 or less. In the organoleptic evaluation of the HTST fluid sterile milk samples, the particular defect or defects that were incurred were recorded as well as the numerical score. RESULTS AND DISCUSSION Organoleptic Flavor Evaluation Organoleptic flavor evaluation of the HTST fluid sterile milk samples was employed throughout the storage study, and provided sensory evaluation of deteriorative changes as the flavor of the fluid sterile milk progressed from fresh to stale or stale-oxidized. As can be observed from Tables 2, 3, and 4 all samples exhibited a highly cooked flavor immediately after heat processing; however, this cooked flavor disappeared, or was masked by the appearance of other off-flavors during subsequent storage. A slight sweet or carmel flavor was also present in the freshly processed milk, as was a distinct coconut flavor. These cooked, carmel, and coconut flavors are attributable to the stringent heating conditions imposed upon the milk to effect sterilization. Harland and Ashworth (1945), Harland et a1. (1955), Hutton and Patton (1952), Rowland (1937), and Larson and Jenness (1950) have shown the cooked flavor to arise from the inter— action of sulfhydryl groups, freed by heat denaturation of the serum protein B—lactoglobulin, and to a minor extent, from the fat globule membrane protein. The interaction of these reactive SH groups forms volatile sulfides, mercaptans, and hydrogen sulfide, all of which are related to cooked flavor in milk. The coconut flavor present in the milk has 41 42 been demonstrated by Keeney and Patton (1956b), Mattick et a1. (1959), Tharp and Patton (1960), and Parliment et a1. (1966) to arise from delta—lactones, which according to the Boldingh and Taylor theory originate from heat degradation of monohydroxy—acyl-triglycerides. The monohydroxy—acyl-tri— glycerides responsible for delta—lactones are those containing a 5-hydroxy fatty acid and two nonhydroxylated fatty acids. Parliment et a1. (1966) have shown that application of heat to these monohydroxylated-acy1-triglycerides results in the hydrolysis of some 5-hydroxy fatty acids, which readily lactonize to form delta—lactones. The sweet, or carmel flavor that also accompanies heated milk products, arises from the heat degradation of lactose, which Kass and Palmer (1940) describe as the carmeli— zation.oflactose by the casein, and the absorption of the lactocarmel by the colloidal caseinates. Concomitant with the carmel flavor in the HTST fluid sterile milk is the off—white to tan color of the freshly processed milk. This discoloration became more apparent with storage of the milk at 22° and 36°C., while the samples at 4.4°C. exhibited no additional color change during storage. The slight discoloration observed with the control (zero time storage) samples is most likely the result of Maillard browning. Patton (1952) observed that browning was readily evident in milk samples heated to 120°C. for 7.5 minutes, and with the heating conditions imposed upon the HTST fluid sterile milk (140.5°C : 1°C. for approximately 4 seconds) it is 43 postulated that Maillard browning results with the consequent formation of melanoidins. The discoloration observed in all control samples as well as the carmelized flavor in one of the control samples, appears to lend support to the theory that sulfhydryl groups are very labile, and disappear readily either as volatile sulfides, or by disulfide interchange with kappa casein (Hartman and Swanson, 1965). The basis for this reasoning lies in the sulfhydryl inhibition mechanism, ob- served by Tarassuk and Jack (1948) and Patton and Josephson (1949b), for heat-induced browning. Hodge (1953) speculated that the inhibition mechanism concerned the addition of sulfhydryl compounds at the double bond of the Amadori rearrange— ment product (in the enol form), effectively blocking the heat— induced browning reaction in some systems. Thus if sufficient reactive sulfhydryls had been present to inhibit Maillard browning, discoloration should not have occurred. Stale flavor development in the HTST fluid sterile milk is the major obstacle associated with flavor deteriora- tion in storage. Until stale flavor was detected organolepti— cally in the stored fluid sterile milk samples a minimum score of 36 on the A.D.S.A.-D.S.I.A. scoring system was recorded for the storage samples. This would be a marketable product. Appearance of the stale flavor lowered the sterile milk score to 34, which is indicative of a product of near unsalable quality. The staling of HTST fluid sterile milk manifests itself between the second and third months of storage at 22°C. This same time lapse was recorded by Patel et a1. (1962) for 44 staling of sterile concentrated milk held at 2l.l°C. It is also evident from Tables 2, 3, and 4 that the storage tempera- ture to which the product is subjected plays an important role in the staling of storage milk products. Here we observe that two out of three samples of HTST fluid sterile milk held at 36°C. were judged slightly stale in their first month of storage. However, pronounced stale flavor was exhibited by the HTST fluid sterile milk somewhere between the first and second months of storage at 36°C. These specific data and the general aging of all sterile milk samples agreed well with the gas chromatographic survey data (Figures 1-10). Here the appearance and enhancement of stale flavor can be correlated with the increased number of carbonyls present as staling develops, as well as the increased concentration of those previously detected in the control samples. Such gas chromatographic data are relevant to stale flavor in stored dairy products, since Bassette and Keeney (1960) have reported the association of autoxidation and Maillard browning to stale flavor, and have corroborated the catalytic effects of high storage temperatures on both autoxidation and Maillard browning reactions. The beneficial effect of refrigerated storage as an inhibitor on staling of HTST fluid sterile milk cannot be concluded from these storage data, although research has pre- sented evidence that lower storage temperatures reduced the intensity of the off-flavors in the sterile milk. Patton (1955), in a review of Maillard browning in dry milk powders, 45 ranks storage temperature second only to moisture content in factors contributing to Maillard browning. Badings (1960) likewise, feels storage temperatures are critical in autoxida— tion of milk lipids. Although the products are in a different physical state, it is apparent that elevated storage tempera— tures are an important factor in contributing to the stale flavor of HTST fluid sterile milk. This view is supported by Patton (1955), who suggests that the lowest feasible storage temperature should be employed in order that flavor, as well as color changes, will be inhibited. Body Defects Associated with HTST Fluid Sterile Milk During Storage at 4.40, 22°, and 36°C. The physical defects that were observed in HTST fluid sterile milk during the storage conditions of refrigerated, ambient, and tropic temperatures are those that have been commonly associated with sterile concentrated and evaporated milks during storage. Cream layer formation was the first body defect that was detected during the storage of HTST fluid sterile milk. This physical separation of the milk fat from the serum phase is explained by the inverse of Stoke's equation for falling bodies, which states that fat globules rise in milk systems for two reasons: a) a density differential exists between milk fat and milk serum, and b) the size of the fat globules. It was observed that even with homogenization pressure of 4000 psig. (3000 psig. lst stage, and 1000 psig. 2nd stage) the fluid sterile milk still developed a cream layer after one month of storage at Table 2. Organoleptic evaluation data for HTST fluid sterile milk under storage conditions of 4.4, 46 months time interval (processing date 3/2/66). 22 and 36°C. for a three Storage Time 4.4°C. 22°C 36°C. Control highly cooked . 35 (zero tlme) coconut One Month 51. cooked 37 cooked 37 51. coconut 36 51. coconut sl. coconut sulfide flavor Two Months lacks fresh. lacks fresh. stale 36 36 34 cooked sl. cooked Three Months astringent stale no sample 34 . 34 stale sl. astrln- gent Table 3. Organoleptic evaluation data for HTST fluid sterile milk under storage conditions of 4.4, months time interval (processing data 5/20/66). 22 and 36°C. for a three Storage Time 4.4°C. 22°C. 36°C. Control highly cooked 37 (zero time) One Month 51. cooked 37 51. cooked 37 lacks fresh. 34 5 31. carmel. sl. carmel. sl. stale ° Two Months sl. cooked lacks fresh. stale sl. coconut 36 51. carmel 36 $1. carmel 34 lacks fresh. sl. coconut Three Months 51. carmel 34 stale 34 stale 34 stale *Organoleptic evaluations were made D.S.I.A. point scoring system. using the A.D.S.A.- 47 Table 4. Organoleptic evaluation data for HTST fluid sterile milk under storage conditions of 4.4, 22 and 36°C. for a three months time interval (processing date 6/9/66). Storage Time 4.400. 22°C. 36°C. Control highly cooked . 36.5 (zero time) 51. carmel. One Month cooked 36 $1. cooked 36 lacks fresh. 34 5 lacks fresh. lacks fresh. Sl. stale ° Two Months lacks fresh. sl. carmel. sl. carmel sl. carmel 36 $1. cooked 36 $1. stale 34 $1. cooked lacks fresh. Three Months lacks fresh. lacks fresh. stale 34 $1. cooked 36 $1. stale 34 $1. carmel sl. cooked *Organoleptic evaluations were made using the A.D.S.A.- .D.S.I.A. point scoring system. 48 temperatures of 22°C. and 36°C.; however, sterile fluid milk held at 4.4°C. did not show any noticeable cream layer after three months of storage. Notation should be made that these samples were not subjected to any type of stock rotation or inversion during storage. The question of what causes the clustering of these rising fat globules, and subsequent formation of the cream layer has never fully been determined. Dunkley and Sommer (1944) established that a protein is involved in this phenome— non, and have classified it as a euglobulin: however, they were not able to establish whether this protein was identical with the euglobulin of immune proteins. The exact effect that the euglobulin has on milk fat is not known. Pre- sumably the protein alters the surface properties of the fat globules in a way which allows them to adhere to one another. It is also known that heating above minimal pasteuri- zation temperature reduces the cream layer volume due to heat denaturation of the euglobulin. Homogenization may also have a similar effect. Explanation of this cream layer or cream plug forma- tion in the storage fluid sterile milk may be attributed to the high homogenization pressure. This has been shown as the cause of cream plug formation in half and half and other homogenized cream products. The euglobulin is not believed to be responsible since it would supposedly be denatured by both sterilization and homogenization (Patton and Jenness, 1959). The observation that creaming occurs in 49 samples stored at 22° and 36°C., but not at 4.4°C. is also intriguing. This seems inconsistent with the early litera- ture which reported that optimal conditions of temperature for creaming of fat in pasteurized products was l.7° to 4.4°C. This temperature range would also show a greater density differential between the serum and fat phases of the milk system, than at any of the other storage temperatures. Age-thickening, an incipient sign of gelation, was first tactually observed in the samples stored for two months at 22° and 36°C., and in the sample held at 4.4°C. for three months. The body of these fluid sterile milk samples could be compared to that of half and half, a dairy product con- taining at least 10% milk fat. This increase in viscosity is believed due to gelation, since no apparent increase in viscosity was detected by organoleptic observations of the sterilized milk samples, and would seem to discount heat coagulation during sterilization as the main cause of increased viscosity. Gelation of sterilized milk appears to be a very complex phenomenon associated with the formation of a fat/ casein complex, which has been shown to be influenced by the following variables: a) heat treatment given the milk, b) composition of the milk, c) concentration of total solids, and d) storage temperatures. The gelation phenome- non is directly related to the temperature of storage. Ellertson and Pearce (1964) have shown that at 4.4°C. con- centrated sterile milk of 26% total solids had a gelation 50 time of 50 weeks, while gelation occurred in less than ten weeks at 36°C. Gelation seems to originate with heat processing, which causes the formation of large fat protein particles. The actual appearance of the gel structure does not begin until these large particles dissociated and rejoin to form short chains. Researchers have suggested that the bonding sites involved in gel structure are not available until the large protein particles dissociate. The gelation sequence has been studied by following the viscosity of milk from a time interval prior to processing until gel formation occurred in the concentrated product. Reduced viscosity was associated with the early stage of dissociation of large particles; these large particles, then dissociated to form smaller particles and short chains, which increased viscosity, and ultimately, a gel structure formed. Further studies on gelled milks by Sasago et a1. (1963) seem to indicate the involvement of disulfide bonds in the binding of the fat globules with the protein, and the importance of calcium in the formation of the large particles. Many researchers have shown that increased processing time and temperature reduces gelation by protein stabilization: however, with this increased heat treatment the problem of sedimentation during storage arises. Phosphates have also been utilized as a means of controlling gelation by complexing the casein micelles with polyvalent cations, which prevents fat-casein complexes from forming. The polyphosphate method 51 of retarding gelation has proven quite effective, as ShOWn by Leviton and Pallansch (1962), and Leviton et a1. (1963), and has increased storage life of concentrated milk from 45 days (control) to 441 days (polyphosphates added). The exact mechanism involved in the retardation of gelation by polyphOSphates is not known. PolyphOSphates have a stabiliz— ing effect on proteins and this may be involved in some way with the reactive sites of the fat globules. Ellertson and Pearce (1964) have shown that homogenization at elevated temperatures increases the viscosity of concentrated sterile milk and this increases the rate of gelation but it is not known whether or not this is due to the effect of homogeni- zation on the casein micelles. Sedimentation was also observed in the samples held for one month. The amount of sediment seemed to be constant in all samples and did not appear to be affected by storage temperature. This body defect is viewed as resulting from the effect of the ultra-high temperature of processing on the milk proteins. In summary, it is felt that after three months of storage at 4.4°, 22°, and 36°C. only the cream layer formation in samples held at 22° and 36°C. was of significance, since normal agitation could not produce sufficient distribution of the two phases. The quantity of sediment could be con— sidered insignificant and the age—thickening could be con— trolled by the addition of polyphosphates. 52 Parameters Incurred in Flavor Recovery Distillation, Sampling, and Gas Chromatography The low-temperature, reduced-pressure distillation technique employed in the flavor volatile recovery system utilized in this research had two important parameters: a) pressure in the distillation flask, and, b) temperature of the distillation. Accurate control over this temperature— pressure combination was extremely important in the standardi- zation of the flavor volatile recovery. This enables the researcher to compare data from different extractions with assurance that either an increase, or decrease in volatiles recovered from the fluid sterile milk was the result of flavor improvement or deterioration, rather than an incon- sistency in the recovery technique. When gas chromatographic analyses of the flavor isolates were initiated they were performed only on the liquid nitrogen cold trap. The reason for this procedure is that prior analyses of each individual cold trap,showed no additional flavor compounds that were not detected in the final liquid nitrogen trap. Therefore, to minimize the number of volatile samples that would have to be chroma- tographed only the flavor isolate from the liquid nitrogen trap was analyzed. As an enrichment procedure the volatile from each of the other cold traps was salted out and the volatiles trans— ferred to the liquid nitrogen trap by manipulating the pressure and temperature on the cold traps. This was done 53 in sequential order from the wet ice to the final dry ice ethanol trap. Following this standardized volatile enrich- ment procedure the flavor isolate was held in a cold trap in liquid nitrogen.(4l95.8°C.) until gas chromatographic analyses were performed. When the flavor isolate held in the liquid nitrogen cold trap was to be prepared for gas chromatographic analysis, the standard taper glass stopper was removed and a rubber stopper, covered with a double layer of parafilm, and fitted with a rubber septum, was inserted in its place. The cold trap was then taken out of the liquid nitrogen bath and slowly warmed to 70°C. When the trap containing the isolate reached equilibrium with the 70°C. bath a 6 ml. sample of isolate volatiles was removed with a gas-tight syringe and injected into the gas chromatograph. During the preliminary trials concerning the gas chromatographic analyses of the flavor isolate vapor an interesting observation was made concerning the sampling of the vapor. It was found that if the vapor sample was removed from the cold finger without releasing the existing pressure from the system only peaks 1 to 8 were chromatographed. On the other hand if this pressure was released and the trap permitted to remain at equilibrium temperature (70°C.) for a few minutes the normal flavor volatile spectrum was obtained. The reason postulated for this behavior concerns the vapor pressures of the low boiling volatiles present in the flavor isolate. It is believed that the low boiling volatiles build 54 up such pressure in the system that the vapor pressure of the higher boiling volatiles is suppressed, and consequently cannot reach the vapor state. In substantiating this theory it was observed that when systems which showed only peaks 1 to 8 had the excess pressure released and were rechromato- graphed, the normal volatile spectrum was obtained. Another experiment that was performed to show this effect was the addition of a wide boiling point range of carbonyls to a cold finger containing distilled water. The trap was then heated to 70°C and a 6 ml. vapor sample was removed and gas chromatographically analyzed. Here again it was observed that until the pressure on the system was released only the low boiling volatiles were being sampled. Because of these findings it was adopted as standard procedure to release the excess pressure on the isolate trap before vapor samples were taken. Previous research also indicated that no volatile components were lost during this procedure. Another observation made concerning the gas chromato- graphic analyses of the flavor isolate volatiles concerns functional group analysis. It was apparent that the functional group reagents that were employed did not always react quanti— tatively with the flavor components. This meant that in some cases the peaks involved were greatly reduced or eliminated while at other times the diminution was only slight. Possibly only the vapor that came into immediate contact with the functional group reagent was chemically treated, which limits the reliability of such a functional group classifi- cation. 55 Identification of Flavor Compounds Detected In HTST Fluid Sterile Hulk by Gas Chromatography The compilation of data interpreted from over 240 gas chromatographic analyses of sterile fluid milk is shown in Table 5. These data relate to the off-flavor development in fluid sterile milk during storage at 4.4°, 22°, and 36°C. In order that the absolute retention time data for each peak could be used as an identification constant, the average retention time of each peak was calculated and is recorded in Table 5. Some compounds isolated in the flavor distillate were found to have identical or nearly identical retention times with one another. This necessitated functional group analyses on the vapor of flavor isolate in conjunction with gas chromatographic analysis, in order to determine in which functional group the unknown compound belonged. Retention times of the proper class of standards could then be compared to the unknown peak for the purpose of tentative identification. The functional groups that were of interest in this research were aldehyde, ketone, alcohol, ester, and sulfide. The technique used in this functional group analysis procedure was that of Hoff and Feit (1964), which has been previously discussed in the procedure. Table 5 represents the compila— tion of the data found in Figures 1—10, which was agas chromatographic survey of fluid sterile milk through a three month storage period at temperatures of 4.4°, 22°, and 36°C. The data in the four columns on the right in Table 4 enable the reader to determine the peaks that were present in the 56 control and storage samples, and those that appeared or disappeared during the storage life of the HTST fluid sterile milk. This information also indicates if any particular peak was present at all storage temperatures, or at only one or two of the storage temperatures. Peak number 1 through 5 were not conclusively identi— fied; however, peaks 1 and 3 appeared to be aldehydes because of their marked response to both of the carbonyl reagents. The functional groups of the components in peaks 3', 5 and 5' could not be determined with sufficient consistency to give any useful information for peak identification. Peak 6 exhibited a retention time of 2.0 minutes, which corresponded to that of ethanal (1.9 min.). The retention time coupled with gthe information gained through functional group analyses suggested the identification of peak 6 as the C2 alkanal. The presence of ethanal in the flavor isolate of fluid sterile milk has also been supported by thin-layer identification of its 2,4 dinitrophenylhydrazone, (Fig. 13). Peak 6', which had an average retention time of 2.8 minutes, has been tentatively identified as propanal. Peak number 7 has not been identified, but functional group analyses classified this component as an aldehyde. Since this peak did not correspond in retention time data to that of any low saturated aldehydes, it is possible that it belongs to the class of unsaturated alkanals. Peak 8 has been tentatively identified as acetone. The presence of acetone in the flavor isolate has also been shown by thin layer separation of its 57 2,4 dinitrophenylhydrazone (Figs. l6, 17). Peak 8' with a retention time of 4.2 minutes was confirmed as an aldehyde, but confirmation as to the carbon number was not obtained. It is interesting that butanal, whose presence in the flavor isolate was also shown by thin-layer separation of its hydra- zone derivative, exhibits a retention time of 4.6 minutes. Because of these data it is believed that peak 8' is the C4n-alkanal. Peak 9 was found to contain two components by functional group analyses. Both ethanol and butanone with identical retention times of 5.6 minutes on the Carbowax »20 M column were shown to contribute to the reSponse of peak 9. This conclusion was drawn after functional group reactions with nitrite and sodium hydroxide showed ethanol present, but failed sufficiently to diminish the peak. This seemed to indicate the presence of another component. Further investi- gation with functional group reagents showed another component present, which was identified as 2—butanone. Peak 10 was not identified and could not be classified by functional group analyses. Peak 10' was functionally classified as an aldehyde, but did not correspond in retention time to any of the alkanal standards. Peak 11 was not identified, but was believed to contain both an alkanal and an alkanone. The reasoning behind this postulation is that during the course of the three storage studies, this peak would some- times react with potassium permanganate, sometimes with acid hydroxyl amine, and sometimes with both reagents. Peak 11' was identified as 2-pentanone. It was not present in the 58 control samples, but appeared to emanate from peak 11 during the first month of storage at 4.4°, and 22°C. This could account for the positive alkanone reaction that was recorded for peak 11. Peak 12 was not tentatively identified, but was functionally classified as a ketone, while peak 12', which was also unidentifiable, was classified as an aldehyde. Peak 13' was believed to contain either or both classes of carbonyls, as indicated by the group reagents. Functional group analyses and the retbntion time data of the known standards 2-hexanone and n-hexanal, served as the basis of this tentative identification. Hexanal was also identified as a 2,4 DNP hydrazone derivative in the flavor isolate of the 3 months 22°C. storage samples by TLC (Fig. 15). Peak 14 was shown as a possible C alcohol by correlation with re- 3 tention data for alcohol standards, and peak diminution when reacted with the nitrite reagent. Peak 14' was tentatively identified as n—butanol, not only because of its comparative retention time with a standard, and decreased peak height when treated with sodium nitrite; but also because of the increased peak height when the vapor sample was treated for the identification of esters with sodium hydroxide. Peaks 14, 15, and 16 were not identified, nor were any functional group analyses data obtained. Peak 16' with a retention time of 23.4 minutes was tentatively identified as heptanal. Peak 17 was identified as 2-heptanone; the C7 alkanone has also been identified by thin-layer chromatography of its 2,4 dinitrophenylhydrazone derivative in the three months 59 22°C. storage samples. Peaks 18, 18', 18", 19 and 20 were not able to be identified. Peak 21 was shown to exhibit the same retention time as n-hexanol but tentative identification was not conferred upon this peak, since it did not consistently react with the functional group reagent employed for alcohol confirmation. Peak 22 was not tentatively identified, but it was functionally classified as an aldehyde. Peak 23 which exhibited the characteristic retention-time and functional group data of 2-nonanone was unable to be confirmed. The only other peak which was tentatively identified gas chromato— graphically was peak 24, which was believed to be furfural. Of the remaining peaks 25 to 30 the only information that could be deduced from the chromatograms was that peaks 27 and 29-30 were most probably aldehydes and peak 26 a ketone. In the following section the components that have been gas chromatographically identified from the flavor isolate, and are believed to be of flavor significance, are discussed. Flavor Vblatiles Tentatively Identified in HTST Fluid Sterile Milk by Gas Chromatography and their Flavor Significance Ethanal Peak number 6, which has a retention time value of 2.0 minutes, has been tentatively identified as ethanal. Ethanal was found in the control and storage samples and as can be observed from the chromatograms (Figs. 1-10) the ethanal concentration remained quite stable from the control 6O Auoomuoe.ev+ + + + excess: o.am as + AooomI0m~v+ -Auoemuoe.ev+ + ezocxas m.om ma Aooe.ev+ I I I sauces: e.om =da + AoommI0mmv+ + + Hocmpsm N.ma .da + AooomI0mmc+ + + Hosmooum e.sa ea + + + + eamcmxmm Iwcocmxmm N.©H .ma + + + + csocxcn ©.NH ma + AOOoMI0mmv+ + + csocxcb o.HH .NH + + + c3ocxso ©.m NH Aooomv+ .AOONNV+ + I emcocmucmm m.m .HH AUONNIoe.dV+ + + + tsetse: o.m as AUONNIoe.eV+ xoonIoe.ev+ Aoemuoe.dc+ + neonate m.s .oe + AuoomI0mmv+ + + csochD m.o 0H + + + + Aw.mvlo.m «msosmusm Iaosmnum m.m o AommI0mmV+ Aoommv+ Aooomv+ I o d eemsmosm m.d .m + + + + v m emcoumom m.m m + + + + s3ocxcb N.m h Aooe.ev+ + + + a m Headache m.m .o. + + + + m H samcmnum o.m o loommv+ 160mmv+ + + saosxso m.H .m + + + + csocxcb d.H m I I + I csochD o.H .m + + + + :3ochD m. m I I I I c3osxnb n. .H + + + + csocxop m . a mmmnoum mmmnoum mmmmoum mxmmm mo .CHS .oz menses m mausoz m Sago: H muaucpr mama xmmm . coflucmpmm m>apmucoa .uom .mnmmnmoumeonnu mom ma >o>n5m ommuoum sucoE mounu m mcHHSU Mame maflnmum UHSHM swam CH pmuomump mpcmcomEoo Ho>mam mo mmEHp coaucmumm .m magma 61 .mm>flum>aump wsoumnpacamcmamonuflcap gem Hams“ mo mammumoumaouco Howma Isms“ an xHHE mafluopm UHDHM swam mo oDMHOmH uo>mam may Ehom poamausmpfl mpcsomEOOe + Aooemc+ AUO¢.¢V+ xooe.dc+ + Aooe.ev+ Aooomv+ Aooe.ev+ AUO®MIONNV+ Aooemuoe.ec+ AUOQMIONNV+ + Aooemv+ + + .Aoom~v+ AOONNV+ loom~c+ Aoommv+ AUOOMIONNV+ AUOQMIONNV+ + Aooomv+ Aoommv+ ACCOMIONNV+ + xuoemv+ + Aooemv+ Aoommo+ + AUO¢.¢V+ loommc+ + + + I +-++-+ I + I+-+-++-+ I csochD ssochD c3ochD G3ochD c3ochD Hmnsmusm czocxcb s3ocxcb csochD csocxcb csostD csoaan :3ocst s3ochD csochD smoocmumwm Hmcmummm ¢.m¢ N.m¢ ¢.o¢ o.mm ¢.®m m.flm o.¢m ¢.mm N.Nm m.om N.om w.mN N.mN 0.0N o.©N N.¢N ¢.mm mm mm mm .mm mm mm .mm mm mm HN ON ma .umnfi .mH ma ha .ma 62 .maoamm Houusoo xHHE ofiflumum madam HmH: Eouw mmawumaop uo>maw use mo compwoomaouco wmw < .H ouswflm AmouscflZv mawH n3 'flé I in; ‘: - a 0’0 -« “'3‘: 1 Recorder Response 63 .ood.d um omeODm Luzoa mco Houmm xaaa mafiuoum pwsam Hmam soum moaaumao> Ho>me ecu mo smmuwoumaouco mew < Amanda—“5 as: H .N muswwm w... a... M e, .G n... O. M ~ ”II: M MW .0. I? who. Recorder Re5ponse 64 .Uomm um owmhoum fiudoa oco nouns Rama oaauoum wflsam HmHm scum mmHHumHo> Hopmam ecu mo :mmuwoumaouco mew < Amousawzv mafia .m ouswwm ] Recorder ResPonse 65 .oocm um owmuoum cocoa moo Houmm xHHa oawsoum vwsam HmHm scum moHHumHo> Ho>mHm map «0 somuwoumaouco new 4 $95836 0.5 H. .d ouswfim Recorder Re3ponse 66 .ood.q um ommuoum mnucoa osu Houmw xHHE oHfiHmum pHSHm HmHm Eoum mmHfiumHo> Ho>me ecu mo someODmaouco mew < .m madman $3255 as: E J 3 1 .. I - W 3: mm 1 FR .2 ale. .5. .. s. I. he mm o rum 0 9 mm» 1 : uON Recorder Reaponse 67 msuaos oSu Houmm saws mafiaoum panam HmHm souw moafiumao> Ho>mam one mo rompwoumsopzo mew < .o ounwwm Amoussfi2v mafia .ooNN um omeOum .9 Al a... p .3 j a. . .fl emu. m. .u .u fie . .E = II. 0. 1 as are . h .0 Md. flu 8N“ and. .AW .0. 1 E a”. In I.- I. Recorder ReSponse .ooom um ommuoum mausos osu Houmm sags oafluoum cwsfim Ewan scum moawumao> uoamam on“ mo newswouuaouno mum < .n onswwm 68 AmmussHZU mafia ail: ( s. 8““ ’L \M. a. .1L .. 4 mm. r .nlu. : a a 5h 1.. s”. ' a sole .u m a. can . e. G noon. H. . u. 1 rd mu..- ct .e. x.” :I: Recorder Response 69 .ood.¢ um owuHOum ocuaoa mousu Houmm xafia meumum pHsHm Hwy: Scum moaauwao> Ho>mam exp mo somnwoumaouao new < $345.36 use; ‘0. w: a ..I.. z a. a: D :. 2 2|: PI I r 9d 4: I: u 3' I: 1' he ~E .m ousmam i] Recorder Response 7O .UoNN um owwuoum mLDGOE moan“ nmumm xHHE mafiumum cHsHm Ewan scum mmHHumHo> uo>mHm or“ mo comnwoumaonao mew 4 .m ouswflm Ammuscfimo mEHH :- JW I L .D n. 1 .m .04 l .E n. .3. 1...: .N. a goon. I In: In” .0. Recorder Reaponse 71 . . .ooom um omMHOum mnucos douse Houmm Mafia oaauoum cHsHm Hmam scum moHHumHo> Hoamam one mo nonsmoumaouco mum < .OH ouswam $3555 95H. 21* l1ll ..... .... 1 a 1 EH Recorder Re3ponse 72 samples through the first month of storage at all three storage temperatures. Between the first and second month of storage the concentration of ethanal increased. This increase was detected in both the second and third month storage samples with the largest concentrations being present at 4.4°C. The reason for the higher alkanal concentration being found in the refrigerated samples could be accounted for in two ways: a) that higher storage temperatures may favor aldolizationcnfthis active carbonyl, and/or b) since ethanal is only slightly less reactive than normal sugar fragentation products in forming carbonyl-amine complexes either or both of these mechanisms could account for the increased browning in the storage samples held at'22° and 36°C. Research has shown that amino compounds, including peptones and albumins do catalyze aldol condensation (Hodge, 1953), such higher alkanals could then more readily react with free amino groups present to form the carbonyl-amino complexes. The contribution of ethanal to the stale flavor that develops in the storage fluid sterile milk is uncertain, since ethanal has also been identified as a flaVor component of raw milk. Propanal Propanal has been tentatively identified as peak 6' in the gas chromatographic analyses of the fluid sterile milk. Propanal was observed in the sterile milk throughout its storage life, and its concentration appeared to be 73 relatively constant until the third month at 22°and 36°C. At this time peak 6' disappeared from the chromatograms. The relatively constant concentration of propanal in the sterile fluid milk would seem to indicate that it may have resulted from the following two sources: a) hydrolysis of lower aldehydes from plasmalogens by sterilization and/or b) it may have been present in the milk before processing. Propanal has also been shown to arise from the autoxidation of linolenic acid. Acetone Acetone, a well known flavor component of dairy products, was tentatively identified by functional group analyses and retention time data. Acetone was also identi— fied in both the storage and control samples by thin-layer chromatography of its 2,4 dinitrophenylhydrazone (Figs. 16. 17). The concentration of acetone in the control sample was quite high, due possibly to the hydrolysis of the B-keto ester or the decarboxylation of B—keto acids during sterili- zation of the milk. Subsequent storage of the fluid sterile milk for one month at 4.4°C. showed a decrease in acetone concentration: this loss was also noted in the one month 22°C. and the two month 4.4°C. and 22°C. storage samples. Increases in the acetone concentration were observed in the two months 36°C. and three months 22° and 36°C. fluid sterile milk samples, which had been judged stale. An increase of 30-50% in acetone concentration in the stale sterile milk 74 samples was noted, depending upon the storage temperature. These increases were most probably caused by storage at elevated temperatures, which previous research has shown to cause Spontaneous decarboxylation of B-keto acids resulting from the B-oxidation of fatty acids. Both Wong and Patton (1962) and Boldingh and Taylor (1962) have presented evidence concerning the ease with which this decarboxylation reaction occurs. If this supposition is true it appears as though the limiting reaction in the formation of acetone could be the B-oxidation of the fatty acids. Butanal The tentative identification of butanal by gas chromatography was supported by thin—layer chromatographic separation of its 2,4 dinitrophenylhydrazone (Fig. 15). Butanal was detected from the following storage samples: one month at 36°C., two months at 36°C., and three months at ° and 36°C. The concentration of butanal detected from 22 the storage samples increased quite noticeably between the first and second months of storage at 36°C., and then leveled off between the second and third months. The fact that butanal was not isolated from the control samples, as a hydrazone derivative, appears to indicate that it is not a normal flavor component of fluid milk before heat processing, or that it arose as a direct result of heat processing. This evidence suggest that the butanal probably originated as a result of lipid autoxidation. This could explain why the 75 butanal was initially found at low concentrations. Then in later storage, as the autoxidative chain reactions progressed, the concentration of butanal increased. Ethanol Ethanol was the first alcohol in the homologous series that was tentatively identified in the gas chromato- graphic effluent. This identification was difficult, since functional group analyses indicated that peak 9 was a compo- site of two flavor compounds; ethanol and 2—butanone. Identi- fication of ethanol was made on the basis of retention time. and its reaction with two functional group reagents. Peak 9 exhibited the normal decrease in recorder response when the flavor isolate vapor was treated with nitrite solution, which is used for alcohol confirmation. Another interesting change in the chromatogram occurred when the effluent coinciding with this peak was treated with sodium hydroxide. Under these alkaline conditions an increase in recorder response was noted. This was believed to indicate not only the presence of ethanol, but an increase in ethanol due to the hydrolysis of an ethyl ester or esters present in the flavor isolate. Peak 9, which contains both ethanol and 2-butanone. was present in all control and storage samples and increased significantly in concentration in the 36°C. storage samples at one and two months, and in the 220 and 36°C. three month storage samples. Because of the binary nature of this peak it was not possible to positively attribute the increase in 76 concentration to either of the components; however, after observing that the peak increased only under elevated storage temperatures it seems likely that the increase was probably due to the methyl ketone present. The exact role played by ethanol in the flavor of dairy products is not fully understood, but Day et a1. (1957) and Cobb et a1. (1963) and others have shown its presence in connection with oxidized and carmel flavors in milk products. Both of these flavors have been implicated with the stale flavor exhibited in stored dairy products. Butanone The second component of peak 9 has been tentatively identified as 2-butanone, a C4 alkanone, whose presence was also confirmed by thin-layer separation of its 2,4 dinitrophenyl— hydrazone (Fig. 16). Because of the ease with which methyl ketones can be formed in heated milk systems, it is believed responsible for the increase in peak 9 at elevated storage temperatures. Hexanal and Hexanone Peak 13' was tentatively identified and shown to contain two components, both carbonyl in nature. Storage increased the concentration of peak 13' with more definite increases being observed in the 36°C. storage samples. The initial presence of peak 13' was probably due to the presence of 2-hexanone, formed as a result of hydrolysis of the milk fat at the high processing temperature. The author suggests 77 that hexanal does not appear until later in the storage life of the sterile fluid milk, when autoxidation of linoleic acid has been initiated. This may account for the large increase in concentration of peak 13' in the three months 22° and 36°C. storage samples. The presence of hexanal has also been shown by TLC analyses of its 2,4 dinitrophenyl- hydrazone (Fig. 15) which was isolated from the three months 22°C. storage samples, but not found in the control samples. Butanol Tentative identification of peak 14' as butanol was made from retention time data comparisons with standards, and functional group analyses. It was also apparent that the butanol peak increased when the flavor isolate was treated with sodium hydroxide. This increase was believed due to the presence of a butyl ester or ester in the flavor isolate: however, no confirmational data could be obtained to support this supposition. It is also interesting that the butanol concentration increased during product storage at temperatures of 22° and 36°C. These increases could have resulted from lipid autoxidation, by the dismutation of hydroperoxides, which has been shown to yield alcohols. Heptanal Heptanal has been tentatively identified as the flavor component present in peak 16'. Its presence was not detected by gas chromatographic analyses in the control samples, but 78 was observed in all of the storage samples. The concentration of this flavor component was observed to increase during product storage with the most significant increases being in 'the two months 36°C. and three months 22°C. samples. Absence of this flavor volatile in the initial control samples would seem to eliminate processing as the factor causing its develop- ment, and suggests that lipid oxidation is the most prdbable factor accounting for its presence. Heptanone The identification of 2-heptanone as a flavor com- ponent, which appeared during storage of the fluid sterile milk was based on retention time and data from functional group analyses. 2-Heptanone was also identified in the thin- layer separation of the methyl ketones (Fig. 17) isolated from HTST fluid sterile milk held for three months at 22°C. Increases in the concentration of 2—heptanone during storage can be seen from inspection of Figures 1-10 and were more predominant in the storage samples held at elevated tempera— tures. This could indicate that 2-heptanone was relevant to the storage flavor of sterile fluid milk. Furfural Peak 24 has been tentatively identified as furfural. This component was present in both the control and storage samples up to two months. In the three months storage analyses this peak was absent at all temperatures. It should be noted‘that the supposed furfural concentration was 79 shown to increase during the first two months of storage. Furfural is a common product of Maillard browning resulting from sugar fragmentation and dehydration as well as from hydroxymethylfurfural conversion, which can result from heat degradation of sugars. Because of the boiling point dif- ferential between the two compounds, this peak was believed to be furfural and not hydroxymethylfurfural. Hydroxymethyl- furfural has a much lower boiling point (furfural--l61.7°C., hydroxymethylfurfural--110°C.) than furfural and this would have permitted hydroxymethylfurfural to elute from the column prior to furfural. The interesting point concerning this tentatively identified compound is that it is not present in the three months storage samples, which undoubtedly contains more browning products than the previous storage samples. This loss could be explained by the aldol conden- sation of furfural, leading to the formation of the carbonyl- amine complex, and eventual melanoidin formation (Rice et al., 1947). Furfural, diacetyl, and hydroxymethylfurfural on the other hand are usually considered to be rather un— reactive, and because of this their concentrations are some- times used as an index to the degree of Maillard browning that has taken place. In View of these conflicting observa- tions the exact fate of the furfural in the three months storage samples is not known. It may also be possible that the high sterilization temperature could have effected the milk system in such a way as to permit the furfural to react in the carbonyl-amine condensation. This condensation reaction 80 has been shown by researchers to take place with lower carbon chain aldehydes, and keto acids, but at a rate slower than the reactions with normal fragmentation products (Hodge, 1953). The participation of these tentatively identified flavor compounds in the stale, or stale-oxidized flavor which has been associated with stored dairy products both in the fluid and dry state can best be explained in the following manner. Whitney and Tracy (1949), who determined that stale flavor was of greater concentration in the fat phase of dairy products, directed the activities of future researchers to this starting point in determining off-flavor compounds responsible for causing the storage products to elicit stale flavor response. Bassette and Keeney (1960) in study— ing defects in non—fat dry milk powder indicated that both Maillard browning and lipid oxidation were involved. They presented evidence for both odd and even chained aldehydes. as well as for furfural and diacetyl. Forss et al. (1960b) have shown the significance of the lower saturated aldehydes C -C 1 9 recently Nawar et a1. (1963) in working with stale flavor in the oxidized flavor of dairy products. More in dry whole milk powder identified formaldehyde, ethanal, and propanal from two separate flavor extractions, which were designed Specifically to extract stale flavor components. Bingham (1964) in working with stale flavor in 3:1 concen- trate tentatively identified ethanal, propanal, and n-pentanal as the n—alkanals present in the stale sterile concentrate. .‘J. y w— 81 The presence of methyl ketones has been shown both for oxidized and stale flavors. Arnold et al. (1966) have identified 2-heptanone, 2—nonanone, 2-undecanone and 2-tri- decanone, by means of mass spectroscopy, as ketones present in the stale flavor of sterile concentrated milk. Nawar et a1. (1963) identified acetone, 2-butanone, and 2-pentanone as the ketones present in stale dry whole milk powder. These data also agree with the findings of Parks and Patton (1961) who suggested that staleness may be caused by a combination of carbonyl compounds with their respective concentrations determining which flavor will be prominant: stale or oxidized. The former flavor apparently results at a lower carbonyl concentration than the latter. Thus, from what can be determined from previous research, and from the finding of this research, it appears that lower aldehydes and methyl ketones do contribute to the stale flavor. Those compounds which appear during product storage may have a greater contri- bution to this flavor defect. Analysis of 2,4 Dinitrophenylhydrazone Derivatives Tentative identification of carbonyls present in HTST fluid sterile milk flavor distillate was made by gas chromatographic analyses utilizing the techniques of volatile retention time, and functional group analyses. Confirmation of these data was accomplished by thin—layer chromatographic identification of their 2,4 dinitrophenylhydrazone derivatives. Following the formation of the 2,4 dinitrophenylhydrazones 82 by the method outlined in the procedure, the carbonyl deriva- tives were separated according to class utilizing the pro- cedure of Schwartz et al.(l962) (Figs. 14, 15, 16, 17). Examination of the column during the separation of carbonyl derivatives from the control samples showed the presence of a gray methyl ketone band, and a tan saturated aldehyde band. The presence of these two groups was also spectrophotometri- cally confirmed by observing their absorption maxima. Fraction 1 (Fig. 11) from the class separation of hydrazone derivatives was observed to have an absorption maximum at 365 millimicrons, and fraction 2 and 3 (Figs. 12, 13) absorption maxima at 355 millimicrons. These correspond to the absorption maxima of methyl ketones and saturated alde- hydes as published by Schwartz et a1. (1962). Class separation of the 2,4 dinitrophenylhydrazone derivatives from the three months 22°C. storage sample, which had been judged stale, showed the presence of 2,4 dienals (lavender band on the Celite 545-Magnesia column) along with the methyl ketone and saturated aldehyde bands. Here again the presence of methyl ketones and saturated aldehydes was confirmed spectrophotometrically using the Beekman DBG recording spectrophotometer (Figs. 11, 12, 13). It was also observed that in the storage samples the methyl ketones band was a charcoal color rather than gray, signifying increased methyl ketone concentration in the storage samples. Spectrophoto- metric confirmation of the presence of 2,4 dienals was not obtained from any of the three class separations of_hydrazone 83 derivatives from the storage samples because they could not be eluted from the column. Even after two liters of 100% chloroform had been passed over the absorption column the 2,4 dienals were not eluted. Working on the assumption that chloroform possessed insufficient polarity to strip the 2,4 dienals from the absorption column a mixture of 10%, 15%, and 30% v/v methanol to chloroform mixtures were tried, but these likewise failed to elute these compounds. It was interesting that no monounsaturated aldehydes were observed in the class separation of the 2.4 dinitro- ‘75:. 1' mr.u..mnn ‘L 4 ‘1 phenylhydrazones of either the control or storage samples. Thin-layer chromatographic separations of the methyl ketones in the control samples by the procedure described by Badings and Wassink (1963) confirmed the presence of acetone and 2-butanone (Fig. 16). Acetone, 2-butanone, and 2-heptanone were identified in the three months storage samples held at 22°C. (Fig. 17). The absolute importance of acetone and butanone in contributing to stale flavor deterioration in fluid sterile milk is uncertain, since their presence has been detected in fresh unheated, as well as heated milks. Wong et a1. (1958), however, have shown that the acetone concentration in heated milk is markedly increased over that of the un- heated product. Increases in the acetone and 2-butanone concentrations were also gas chromatographically observed during the storage of the fluid sterile milk (Figs. 1-10). 84 2-Heptanone which, is not normally found in unheated milk, has been shown to arise from the decarboxylation of B-keto acids. B-keto acids have two possible precursors in milk fat: a) as intermediates in the 8 oxidation of fatty acids liberated from triglycerides, and/or b) from the hydrolysis of fi-keto esters, which have been shown in milk fat by Van der ven (1963), and Parks et a1. (1964), both of iii? whom have isolated and identified methyl ketone precursors in butterfat as B—keto esters. The important condition required in the formation (ruin-n - — of methyl ketones in processed dairy products, as shown by Langler and Day (1964), is heating of the milk fat in the presence of moisture. In a process such as milk sterili- zation, hydrolysis of both fatty acids and fi—keto esters has been shown to occur. The importance of water in this reaction cannot be overstressed, since inhibition of methyl ketone formation was observed when anhydrous milk fat was heated. In view of this evidence, it seems possible that the presence of 2-butanone in the controls and in the storage samples could also be attributed to heat sterilization of the fluid milk. The appearance of 2-heptanone in the storage samples. (Fig. 17) is believed due to the decarboxylation of B-keto acids during the prolonged storage at elevated temperatures. The importance of methyl ketones in contributing to flavor deterioration in heat processed milk products held in storage, and possibly to the stale flavor has not been 85 completely determined. However, it has been shown that, commercially feasible, inhibition of autoxidation, and associated oxidized flavor does not necessarily solve the stale flavor problem. Nawar et a1. (1963) isolated stale flavor components from dry whole milk powder by Girard-T-reagent, and carbon- tetrachloride distillations, showing the association of acetone, 2-butanone, and 2-pentanone with stale flavor, Arnold et a1. (1966) have confirmed the presence of 2-hepta— none in stale sterile concentrated milk by mass spectroscopy. Day and Lillard (1960) have also identified acetone as a component in oxidized flavor of milk fat, and at the same time have found presumptive evidence for the involvement of odd-numbered C -C 5 15 postulated that methyl ketones in a medium concentration n-alkan-2-ones. Parks and Patton (1961) elicit a stale flavor response, which they believe is due to their synergistic interaction. Spectrophotometric analyses of the second and third fractions (Figs. 12, 13) collected from the column separation of the 2,4 dinitrophenylhydrazones from the control and the three months storage samples at 22°C. showed them to be saturated aldehydes. Acetaldehyde was identified in the control samples as well as the storage samples (Figs. 14, 15) and because of this was believed to originate from a source. or sources, other than lipid autoxidation. Dutra.et a1. (1959) have shown acetaldehyde formation form the enzyme-catalyzed oxidative decarboxylation (Strecker degradation) of alanine, and the breakdown of lactose by the action of heat during the 86 processing of evaporated milks. It was also shown that the ascorbic acid, which is present in milk, appears to have a catalytic effect on the Strecker degradation of alanine, since the reaction was accelerated by addition of ascorbic acid. They also postulated that other amino acids could under- go this degradation, and yield carbonyls of considerably more flavor significance. It can be seen from the gas chromatographic data presented in Figs. l-lO that the acetal- dehyde level did increase during storage at all temperatures although the increases were noticeably less in the samples held at 4.4°C. These increases in acetaldehyde levels were believed due to lipid autoxidation as well as Strecker de— gradation of amino acids. Acetaldehyde has been related to stale flavor by the work of Nawar et a1. (1963), who re- ported acetaldehyde and propanal present in two separate extractions, which were Specifically designed to extract stale flavor components. Two other saturated aldehydes were isolated, and tentatively identified by thin-layer chromatography from the three months storage samples: n-butanal and n-hexanal (Fig. 15). These compounds were also present in the gas chromatographs discussed earlier. Since these two compounds were not isolated as hydrazone derivatives in the control samples, it is believed that they arose as dismutation pro- ducts of hydroperoxides resulting from lipid autoxidation, rather than the Strecker degradation of amino acids. Hexanal has been shown as a secondary reaction product from the 87 oxidation of linoleic and arachidonic acids, which are both present in milk fat. Butanal has also been isolated from autoxidized milk fat by Day and Lillard (1960), and Nawar et a1. (1962); however, as yet no reaction scheme has been postulated to account for its presence. The flavor signifi- cance of these lower saturated aldehydes in oxidized and stale flavors has been proposed by Nawar et a1. (1962), 1 who postulated that the C -C alkanals contribute to the 1 9 oxidized flavor. They are generally present in small concentrations, and function in synergistic interactions to { cause perceptible flavor deterioration. Parks and Patton (1961) in following this line of reasoning found that low aldehyde concentrations appeared to elicit a stale flavor response, and as the concentration was increased the flavor observed was recorded as oxidized. The 2,4 dienals whose presence was shown during the class separation of 2,4 dinitrophenylhydrazones from the storage samples, were believed to arise as the result of autoxidation of linoleic, linolenic and arachidonic fatty acids according to the mechanism of Farmer and Sutton (1943). It was unfortunate that these diunsaturated aldehydes could not be recovered from the column for confirmatory thin-layer chromatography, since there is general agreement among all workers as to the importance of diunsaturated alkenals in both oxidized and stale flavors. It was also observed during these class separations that no monounsaturated alkenals were present in either the control or storage samples. Absorbance Absorbance Absorbance l 1% 1" let an. E“ Wavelength (millimicrons) ‘ Figure 11. Visible absorption spectra of 2,4 DNP hydrazones in chloroform recovered from fraction 1 during column separation of 2,4 DNP hydrazones from sterile milk at 36°C. for three months, and control samples. P W-S.s. '.' 1 1 4 too 1“ Wavelength (millimicrons) Figure 12. Visible absorption spectra of 2,4 DNP hydrazones in chloroform recovered from fraction 2 during column separation of 2,4 DNP hydrazones from sterile milk at 36°C. for three months, and ntrol samples. 1 1 1 too ass : Wavelength (millimicron33 Figure 13. Visible absorption spectra of 2,4 DNP hydrazones in chloroform recovered from fraction 3 during column separation of 2,4 DNP hydrazones from sterile milk at 36°C. for three months, and control samples. 89 Solvent Front Rt 8.68 I", R! 1'60 a..." m:.50 I"- Reg.“ ‘43 Rfl.3l '...' 3:525 ”3.25 R' 2;“ .: 2' in II“: 0 o i o ’0 c2 c3 04 x2 XI 05 06 07 Figure 14. Separation of 2,4 DNP hydrazone derivatives of saturated aldehydes from HTST fluid sterile milk control samples by thin-layer chromatography. Solvent Front emu 'njj 3:2.23 99:.23 :.:,.' 1"“: 1...} C2 C3 C4 X2 XI C5 C6 C7 Figure 15. Separation of 2,4 DNP hydrazone derivatives of saturated aldehydes from HTST fluid sterile milk, held at 22°C. for three months, by thin-layer chromatography. 9O Solvent Front Rf=.80 :1“. m=.1a(?‘: ‘°" 0" ." a.) RLLBA Rf :52 {:2- mfso an.“ a". .20 00:... k.) ‘..I Rf 8.4l Rf 8.“ 3g...) (.3 C5 C4 C3 X C7 CB C9 Figure 16. Separation of 2,4 DNP hydrazone derivatives of methyl ketones from HTST fluid sterile milk control samples by thin-layer chromatography. Solvent Front Rf8.08 “’8'...‘ :0"o. Rf 8:77 l°°| ‘.:j ‘o... O a" ‘91.: ‘~-' III-.77 if: Rf:.54 ":5“ o .0 .‘ '.‘ 3;) '1 ,3 M 3:39 ”5:39 i": 5 '1- Figure 17. Separation of 2,4 DNP hydrazone derivatives of methyl ketones from HTST fluid sterile milk, held at 22°C. for three months, by thin-layer chromatography. SUMMARY AND CONCLUSIONS 1. Fluid whole milk, which was sterilized at 140.5°C._i~ l.°C. for approximately four seconds in a DeLaval vacu- Therm-Instant-Sterilizer, exhibited a highly cooked and slight sweet taste by organoleptic evaluation of the control samples. The color of these control samples was an off- white to tan, signifying the occurrence of Maillard browning during the sterilization process. 2. Fluid sterile milk stored at temperatures of 4.4° and 22°C. for two months still had an acceptable flavor. Stale flavor was organoleptically detected in the product after one month of storage at 36°C., and two months at 22°C. 3. Age-thickening and sedimentation were observed in the fluid sterile milk during storage. Sedimentation occurred in all sterile milk samples after one month of storage. Age-thickening, an incipient sign of gelation, was visually and tactually detected in the two months 22° and 36°C. samples, and by the third month in the 4.4°C. samples. 4. Severe creaming occurred in the fluid sterile milk after one month of storage at 220 and 36°C. After two months of storage at these temperatures a firm cream-plug had formed in the neck of the bottle. This defect was not observed in the sterile milk samples held at 4.4°C. for three months. All samples were held under static storage conditions. 91 92 5. Standard plate counts were made on the storage samples of the fluid sterile milk before flavor distillation to confirm the sterility of the milk samples and to insure that any flavors present in the sterile milk were not the direct result of bacterial contamination of the milk during storage. No organisms were detected in any of the sterile fluid milk storage samples used for extraction of flavor volatiles. 6. Flavor components were recovered from the fluid sterile milk by a low-temperature, reduced-pressure distilla— tion technique. Separation of the flavor components in the flavor distillate was accomplished by gas chromatography using a packed column and a dual flame ionization detector. 7. Tentative identification was made of the flavor components separated gas chromatographically, using the retention time and functional group analysis data. 8. Flavor components that were tentatively identified in the fluid sterile milk control samples were: ethanal, propanal, acetone, 2—butanone, 2-hexanone, butanol, ethanol, and furfural. 9. 2-Pentanone, hexanal, butanal, heptanal. and 2— heptanone were tentatively identified from the fluid sterile milk during storage as well as those previously mentioned as present in the control samples. An increase of 30—50% in acetone concentration was also observed in the samples, which had been organoleptically judged stale. A general increase in volatile concentration was apparent in all of the sterile milk samples held in storage. 93 10. The presence of ethanal, butanal, hexanal, acetone, 2-butanone, and 2—heptanone in the samples of sterile milk was confirmed by thin-layer chromatography. Butanal, hexanal, and 2-heptanone were found to be present only in the storage fluid sterile milk samples. The presence of 2,4 dienals was observed in the storage samples during the class separation of the hydrazone derivatives, but these compounds were not present in the controls. No monounsaturated alkenals appeared to be present either in the control or storage samples. These organic compounds, whose presence has been shown in the fluid sterile milk, are believed to contribute to the stale flavor by their synergistic interactions, especially those which arose during the storage of the sterile milk samples. LITERATURE CITED Arnold, R. G., Libbey, L. M., and Day, E. A. 1966. Identi- fication of components in the stale flavor fraction of sterile concentrated milk. J. Dairy Sci., 31, 566. Bading, H. T. 1960. Principles of autoxidation processes in lipids with special regard to the development of autoxi- dation off—flavors. Neth. Milk and Dairy Journal, 13. 215. Badings, H. T. and Wassink, J. G. 1963. Separation and identification of aliphatic aldehydes and ketones by thin-layer chromatography of the 2,4 dinitrophenyl— hydrazones. Neth. Milk and Dairy Journal, 11, 132. Bassette, R. and Keeney, M. 1960. Identification of some volatile carbonyl compounds fron nonfat dry milk. J. Dairy Sci., 43, 1744. Bassette, R., Ozeris, S., and Whitnah, C. 1962. Gas chromatographic analysis of head space gas of dilute aqueous solutions. Anal. Chem., 34, 1540. Billis, H. E. and Slowinski, E. J., Jr. 1956. Application of vapor chromatography to infrared spectroscopy of liquids. J. of Chem. Phys., gg, 794. Bingham, R. J. 1964. Gas chromatographic studies on the volatiles of sterilized concentrated milk. Ph.D. Thesis. University of Wisconsin. Boldingh, J. and Taylor. R. J. 1962. Trace constituents of butterfat. Nature, 194, 909. Bradley, R. L. and Stine, C. M. 1962. Simple device for obtaining samples of headspace gas directly from sealed containers for analysis by gas chromatography. J. Dairy Sci., 45, 10. Buttery, R. and Teranishi, R. 1961. Gas-liquid chromato- graphy of aroma of vegetables and fruits. Anal. Chem., 3;, 1439. Chang, S. S., Ireland, C. E., and Tai, H. 1961. An infrared gas cell for the direct collection of gas chromato- graphic fractions. Anal. Chem., 3;, 479. 94 95 Cheronis and Entrikin. 1957. Semimicro Qualitative Organic Analysis. Interscience Publishers, Inc., New York. 2nd edition. Cobb, W., Patton, S., and Grill, H. 1963. Occurrence of vanillin in heated milks. J. Dairy Sci., 45, 566. Coffman, J., Smith, D., and Andrews, J. 1960. Analysis of volatile food flavors by gas-liquid chromatography. I. The volatile components from dry blue cheese and dry romano cheese. Food Res., 25, 663. Day, E. A., Forss, D. A., and Patton, S. 1957. Flavor and odor defects of gamma-irradiated skimmilk. I. Preliminary observations and the role of volatile carbonyl compounds. J. Dairy Sci., 49, 922. Day, E. A. and Lillard, D. A. 1960. Autoxidation of milk lipids. I. Identification of volatile monocarbonyl compounds from autoxidized milk fat. J. Dairy Sci., 4;. 585. Day, E. A., Lillard, D. A., and Nontgomery, M. W. 1963. Autoxidation of milk lipids. II. Effect on flavor of the additive interactions of carbonyl compounds at subthreshold concentrations. J. Dairy Sci., 45, 291. Day, E. A. and Lindsay, R. C. 1963. Methyl sulfide and the flavor of butter. J. Dairy Sci., 45, 615. Day, E. A., Lindsay, R. C., and Forss, D. A. 1964. Dimethyl sulfide and the flavor of butter. J. Dairy Sci., 32, 197. Dal Nogare, S. and Juvet, R. 1962. Gas-liqgid Chromatography Theory and Practice. Interscience Publishers, Inc. New York. Dimick, K. P. and Corse, J. 1956. Gas chromatography-A new method for the separation and identification of volatile materials in foods. Food Technology, 59, 360. Dougherty, R. W., Shipe, W., Guananson, G., Ledford, R., Peterson, R., and Scarpellino, R. 1962. Physio- logical mechanisms envolved in transmitting flavors and odors to milk. I. Contribution of eructated gases to milk. J. Dairy Sci., 45, 472. Dunkley, W. L. and Sommer, H. H. 1944. The creaming of milk. Wisconsin Agriculture Experiment Station. Res. Bulletin. 151. 96 Dutra, R. C., Jennings, W. G., and Tarassuk, N. P. 1959. Flavor compounds from commercial evaporated milk. Food Res., g4_688. Ebert, A. A. 1961. Improved sampling and recording system in gas chromatography-time-of-flight mass Spectro— metry. Anal. Chem., 55, 1865. Ellertson, M. E. and Pearce, S. J. 1964. Some observations on the physical—chemical stability of sterile con- centrated milks. J. Dairy Sci., 41, 564. Ellis, G. W. 1950. Autoxidation of the fatty acids. III. The oily products from elaidic and oleic acids. The formation of monoacyl derivatives of dihydroxy- stearic acid and of a, B-unsaturated keto acids. J. Biochem., 45, 129. Ellis, R., Gaddis, A. M., and Currie, G. T. 1958. Paper chromatography of 2,4 dinitrophenylhydrazones of saturated alipatic aldehydes. Anal. Chem., 59, 475. El—Negoumy, A. M., Miles, D. M., and Hammond, E. G. 1961. Partial characterization of the flavors of oxidized butteroil. J.-Dairy Sci., 44, 1047. Farmer, E. H. and Sutton, D. A. 1943. The course of autoxidation reaction in polyisoprenes and allied compounds. IV. The isolation and constitution of photochemically formed methyl oleate peroxide. J. Chem. Soc., 119. Farmer, E. H., Koch, H. P., and Sutton, D. A. .1943. The course of autoxidation reactions in polyisoprene and allied compounds. VII. Rearrangement of double bonds during autoxidation. J. Chem. Soc., 541. Forss, D. A., Pont, E. G., and Stark, W. 1955a. The volatile compounds associated with oxidized flavor in skim milk. J. Dairy Res., 52, 91. Forss, D. A., Pont, E. G., and Stark, W. 1955b. Further observations on the volatile compounds associated with oxidized flavor in skim milk. J. Dairy Res., 2;. 345. Forss, D. A., Dunstone, E. A., and Stark, W. 1960a. Fishy flavor in dairy products. J. Dairy Res., 21, 211. Forss, D. A., Dunstone, E. A., and Stark, W. 1960b. The volatile compounds associated with tallowy and painty flavors in butterfat. J. Dairy Res., 21, 381. 97 Forss, D. A., Ramshaw, E. H., and Stark, W. 11962. Vinyl ketones in oxidized fats. J. Am. Oil Chem. Soc., 32, 308. Forss, D. A. 1964. Fishy flavor in dairy products. J. Dairy Sci., 41, 245. Gaddis, A., Ellis, R. and Currie, G. 1961. Carbonyls in oxidizing fat. V. The composition of neutral mono— carbonyl compounds from autoxidized oleate, lineoleate, lineolenate esters and fats. J. Am. Oil Chem. Soc., fil. 371. Gohlke, R. C. 1959. Time—of-flight mass spectrometry and gas liquid chromatography. Anal. Chem.,-1L. 535. Gould, I. A. and Sommer, H. H. 1939. Effect of heat on milk with Special reference to the cooked flavor. Mich. Agr. Expt. Sta. Tech. Bulletin, 164. Gould, I. A 1945. Lactic acid in dairyproducts. III. The effect of heat on total acid and lactic acid production and on lactose destruction. J. Dairy Sci., 28, 367. Gould, I. A. 1945. The formation of volatile acids in milk by high temperature treatment. J. Dairy Sci., 28, 379. Gould, I. A. and Keeney, P. G. 1957. Certain factors affecting the concentration of active sulfhydryl compounds in heated cream. J. Dairy Sci., 49, 297. Hannan, R. S. and Lea, C. H. 1952. Studies of the reaction between proteins and reducing sugars in the 'dry' state. VI. The reactivity of the terminal amino groups of lysine in model systems. Biochem. and Biophys. Acta., 9, 293. Harland, H. A. and Ashworth, U. S. 1945. The preparation and effect of heat treatment on the whey proteins of milk. J. Dairy Sci., 28, 879. Harland, H. A., Jenness, R., and Coulter, S. T. 1947. Changes produced in milk on heating. J. Dairy Sci., £9, 526. Harland, H. A., Coulter, S. T., Townley, V. H., and Jenness, R. 1955. A quantitative evaluation of changes occurring during heat treatment of skimmilk at temperatures ranging from 1700 to 3000F. J. Dairy Sci., 48, 1199. 98 Hartman, G. H., Jr. and Swanson, A. M. 1965. Changes in mixtures of whey protein and kappa-casein due to heat treatments. J. Dairy Sci., 4g, 1161. Hodge, J. E. 1953. Chemistry of browning reactions in model systems. J. Agr. Food Chem., 1, 928. Hodge, J. E. and Rist, C. E. 1953. The Amadori rearrange- ment under new conditions and its significance for non—enzymatic browning reactions. J. Am. Chem. Soc.. 15, 316. Hoff, J. E. and Feit, E. D. 1964. New technique for function- al group analysis in gas chromatography. Syringe reactions. Anal. Chem., 36, 1002. Hornstein, I., and Crowe, P. 1962. Gas chromatography of food volatiles. Anal. Chem., 44, 1345. Hutton, J. T. and Patton, S. 1952. The origin of sulfhydryl groups in milk proteins and their contributions to "cooked" flavor. J. Dairy Sci., 45, 699. Jackson, H. and Morgan, M. 1954. Identity and origin of the malty aroma substance from milk cultures of g, Lactis var. maltigenes. J. Dairy Sci., 41, 1316. Josephson, D. V. and Doan, F. J. 1939. Cooked flavor in milk, its sources and significance. Milk Dealer, 2.9.. 29- Kass, J. P. and Palmer, L. S. 1940. Browning of autoclaved milk, chemical factors involved. Ind. Eng. Chem., gg. 1360. Kawahara, F. K. and Dutton, H. J. 1952. Volatile cleavage products of autoxidized soybean oil. J. Am. Oil Chem. Soc., 22, 372. Keeney, M. and Doan, F. J. 1951a. Studies on oxidized milk fat. I. Observations on the chemical properties of the volatile flavor material from oxidized milk fat. J. Dairy Sci., 34, 713. Keeney, M. and Doan, F. J. 1951b. Studies on oxidized milk fat. II. Preparation of 2,4 dinitrophenylhydrazones from the volatile material from oxidized milk fat. J. Dairy Sci., 34, 719. Keeney, M. and Doan, F. J. 1951c. Studies of oxidized milk fat. III. Chemical and organoleptic properties of volatile material obtained by fractionation with various solvents and Girard's reagent. J. Dairy Sci., 34, 728. 99 Keeney, P. G. and Patton, S. 1956a. The coconut-like flavor defect of milk fat. I. Isolation of the flavor compound from butter oil and its identification as delta—decalactone. J. Dairy Sci., 22, 1104. Keeney, P. G. and Patton, S. 1956b. The coconut-like flavor defect of milk fat. III. Demonstration of delta- decalactone in dried cream, dry whole milk and evaporated milk. J. Dairy Sci., 22, 1114. Keith, R. and Day, E. A. 1963. Determination of the classes of free monocarbonyl compounds in oxodizing fats and oils. J. Am. Oil Chem. Soc., 42, 121. Kern, J., Weiser, H., Harper, W., and Gould I. A. 1954. Observations on organic acids formed during the heat sterilization of milk. J. Dairy Sci., 21, 904. Kurtz, F. E. 1965. Direct recovery of added ketones from foam-dried whole milk. J. Dairy Sci., 42, 269. Langler, J E. and Day, E. A. 1964. Development and flavor properties of methyl ketones in milk fat. J. Dairy Sci., 41, 1291. Larson, B. L. and Jenness, R. 1950. The reducing capacity of milk as measured by an iodimetric titration. J. Dairy Sci., 22, 896. Lawrence, R. C. 1965. Use of 2,4 dinitrophenylhydrazine for estimation of micro amounts of carbonyls. Nature, 205, 1313. Lea, C. H. 1950. The role of amino acids in the deterioration of food: the browning reaction. Chem. and Ind., 155. Lea, C. H. and Hannan, R. S. 1950. Reactions between proteins and reducing sugars in the dry state. III. Nature of the protein groups reaction. Biochem. and Biophys. Acta., 2, 433. Lea, C. H. and Hannan, R. S. 1950. Biochemical and nutri— tional significance of the reaction between proteins and reducing sugars. Nature, 165, 438. Lea, C. H., Moran, T., and Smith, J. A. 1943. The gas packing and storage of milk powders. J. Dairy Sci., 1}, 162. Leviton, A. and Pallansch, M. L. 1962. High-temperature- short-time sterilized evaporated milk. IV. The retardation of gelation with condensed phosphates, manganous ions, polyhydric compounds, and phOSpha- tides. J. Dairy Sci., 42, 1045. 100 Leviton, A., Anderson, H. A., Vittel, H. E., and Vistal, J. H. 1963. Retardation of gelation in high-temperature- short—time sterilized milk concentrates with poly- phosphates. J. Dairy Sci., 46, 310. Lewis, W. R. and Quackenbush, F. W. 1949. The use of the polargraph to distinguish between the perioxide structures in oxidized fat. J. Am. Oil Chem. Soc., 26, 53. Libbey, L. M. and.Day, E. A. 1963. Reverse phase thin- layer chromatography of 2,4 dinitrophenylhydrazones of n-alkanals and n-alkan-2-ones. J. Chromatography, 24, 273. Libbey, L. M., Bills, D. D. and Day, E. A. 1963. A technique for the study of lipid-soluble food flavor volatiles. J. Food Sci., 22, 329. Lindeman, L P. and Annis, J. L. 1960. Use of a conventional mass spectrometer as a detector for gas chromato- graphy. Anal. Chem., 22, 1742. Lynn, Steele, and Staple. 1956. Separation of 2,4 dinitro- phenylhydrazones of aldehydes and ketones by paper chromatography. Anal. Chem., 22, 132. McFadden, W. H., Teranishi, R., Black, D. R., and Day, J. C. 1963. Use of capillary gas chromatography with a time-of—flight mass spectrometer. J. Food Sci., gg, 316. Mabbitt, L. A. and MbKinnon, G. 1963. The detection of volatile components of milk by gas—liquid chromato— graphy and its possible application in assessing keeping quality and flavor. J. Food Res., 22, 359. Nbckay, D. and Berdick, M. 1961. Objective measurement of odor-ionization detection of food volatiles. Anal. Chem., 22, 1369. Mattick, L. R., Patton, S., and Keeney, P. G. 1959. The coconut-like flavor defect from milk fat. III Observations on the origin of delta-decalactone in fat containing dairy products. J. Dairy Sci., 42, 791. Meneffee, S. G., Overman, O. R., and Tracy, P. H. 1941. The effect of processing on the nitrogen distribution in milk. J. Dairy Sci., 24, 953. 101 Merritt, C. and Walsh, J. 1962. Qualitative gas chromato- graphic analysis by means of retention volume constants. Anal. Chem., 24, 903. Morgan, M. E., Patton, S. and Forss, D. A. 1957. Volatile carbonyl compounds produced in skimmilk by high- heat treatment. J. Dairy Sci., 42, 571. Morgan, M. E. and Day, E. A. 1965. Simple on-column trapping procedure for gas chromatographic analysis of flavor volatiles. J. Dairy Sci., 46, 1382. Ruck, G. A., Tobias, J., and Whitney, R. McL. 1963. Flavor of evaporated milk. I. Identification of some compounds obtained by the petroleum ether solvent partitioning technique from aged evaporated milk. J. Dairy Sci., 46, 774. Nawar, W. and Fagerson, I. 1960. Technique for collection of food volatiles for gas chromatography. Anal. Chem., 22, 1535. Nawar, W., Cancel, L., and Fagerson, I. 1962. Heat- induced changes in milk fat. J. Dairy Sci., 46, 1172. Nawar, W., Lombard, S., Da11, H., Ganguly, A., and Whitney, R. McL. 1963. Fraction of the stale-flavor components of dried whole milk. J. Dairy Sci., 46, 671. Parliment, T. H., Nawar, W. W., and Fagerson, I. S. 1966. Origin of delta-lactones in heated milk fat. J. Dairy Sci., 42, 1109. Parks, 0. W. and Patton, S. 1961. Volatile carbonyl compounds in stored dry whole milk. J.-Dairy Sci., 44, 1. Parks, 0. W., Keeney, M., and Schwartz, D. 1961. Bound aldehydes in butteroil. J. Dairy Sci., 44, 1940. Parks, 0. W., Keeney, M., and Schwartz, D. P. 1963. Carbonyl compounds associated with the off—flavor in Spontaneously oxidized milk. J. Dairy Sci., 46, 295. Parks, 0. W., Keeney, M., and Schwartz, D. P. 1964. Isolation and characterization of the methyl ketones precursors in butter fat. J. of Lipid Res., 6, 232. Patel, T. D., Calbert, H. E., Morgan, D. G., and Strong, F. M. 1962. Change in the volatile flavor components of sterilized concentrated milk during storage. J. Dairy Sci., 46, 601. 102 Patton, S. and Josephson, D. V. 1949a. The isolation of furfuryl alcohol from heated skimmilk. J. Dairy Sci., 22, 222. Patton, S. and Josephson, D. V. 1949b. Observations on the application of the nitroprusside test to heated milk. J. Dairy Sci., 22, 398. Patton, S. 1950. The formation of maltol in certain carbo- hydrate-glycine systems. J. of Biochem., 184, 131. Patton, S. 1950. Studies of heated milk. I. Formation of 5-hydroxy—methy1-2—furfura1. J. Dairy Sci., 22, 324. Patton, S. 1952. Studies of heated milk. IV. Observations on browning. J. Dairy Sci., 26, 1053. Patton, S. 1955. Browning and associated changes in milk and its products: a review. J. Dairy Sci., 22, 457. Patton, S., Day, E. A., Forss, D. A. 1956. Methyl sulfide and the flavor of milk. J. Dairy Sci., 22, 1469. Patton, S., and Flipse, R. J. 1953. Studies of heated milk. V. The reaction of lactose with milk protein as shown by 1actose-1-Cl4. J. Dairy Sci., 26, 766. Patton, S. and Tharp, B. 1959. Formation of methyl ketones from milk fat during steam distillation and saponifi- cation. J. Dairy Sci., 42, 49. Patton, S. and Jenness, R. 1959. Principles of Dairy Chemistry. John Wiley and Sons, Inc. New York. Patton, S. 1961. Gas chromatographic analysis of flavor in processed milk. J. Dairy Sci., 44, 207. Rhoades, J. W. 1958. Sampling method for analysis of coffee volatiles by gas chromatography. J. Food Res., 22, 254. Rice, R. G., Kertesz, Z. I., and Stotz, E. H. 1947. Color formation in furfural systems. J. Am. Chem. Soc., 62, 1798. Rowland, S. J. 1937. Denaturation and degradation of proteins at temperatures of 75—1200C. J. Dairy Res., 2, 1. Rowland, S. J. 1938. The protein distribution in normal and abnormal milk. J. Dairy Res., 2, 47. 103 Sasago, K., Wilson, H., Herreid, E. 1963. Determination of sulfhydryl and disulfide groups in milk by the para-chloromercuribenzoate-dithizone method. J. Dairy Sci., 46, 1348. Schultz, H. W., Day, E. A., and Sinnhuber, R. 0., Editors. 1962. Symposium of Foods ”Lipids and their Oxidation." A. V. I. Publishing Company, Inc. Westport, Connecticut. Schwartz, D. P., Haller, H. S., and Keeney, M. V1963. ‘Direct quantitative isolation of monocarbonyl compounds from fats and oils. Anal. Chem., 26, 2191. Schwartz, D. P and Parks, 0. W. 1961. Preparation of carbonyl-free solvent. Anal. Chem. 22, 1396. Schwartz, D. P., Parks, 0. W., and Keeney, M. 1962. Separation of 2,4 dinitrophenylhydrazone derivatives of aliphatic monocarbonyls into classes on magnesia. Anal. Chem., 24, 669. Schwartz, D. P., Speegler, P. S., and Parks, 0. W. 1965. Effect of water on methyl ketone formation in butter- oil. J. Dairy Sci., 42, 1387. Senn, V. J. 1963. Changes in total carbonyl content of orange juice and concentrated during storage. J. Food Sci., 22, 531. Shipstead, H. and Tarassuk, N. P. 1953. Chemical changes in dehydrated milk during storage. J. Agr. and Food Chem., 2, 613. Shriner, R. L., Fuson, R. C., and Curtin, D. Y. 1956. The Systematic Identification of Organic Compgunds. 4th edition. John Wiley and Sons, Inc. New York. Soukup, R. J., Scarpellino, R. J., and Danielczik, Ellen. 1964. Gas chromatographic separation of 2,4 dinitro- phenylhydrazone derivatives of carbonyl compounds. Anal. Chem., 26, 2255. Stark, W. and Forss, D. A. 1964. A compound responsible for mushroom flavor in dairy products. J. Dairy Res., 2;, 253. Stull, J. W. 1953. The effect of light on activated flavor development and on the constituents of milk and its products: a review. J. Dairy Sci., 26, 1153. Supplee, G. C. 1926. Humidity equilibria of milk powders. J. Dairy Sci., 2, 50. 104 Swern, D., Coleman, J. E., Knight, H. B., Ricciuti, C., Willitis, C. O., and Eddy, C. R. 1953. Reactions of fatty materials with oxygen. XIV. Polarographic and infrared spectrophotometric investigation of peroxides from autoxidized methyl oleate. J. Am. Chem. Soc., 16, 3135. Tarassuk, N. P. and Jack, E. L. 1948. A study of the browning reaction in whole milk powder and ice cream mix powder. J. Dairy Sci., 22, 255. Tharp, B. and Patton, S. 1960. Coconut-like flavor defect of milk fat. IV. Demonstration of delta-dodecalactone in the steam distillate from milk fat. J. Dairy Sci., 3;: 475. Toan, Tran T., Bassette, R., and Claydon, T. J. 1965. Methyl sulfide production by Aerobacter aerogenes in milk. J. Dairy Sci., 46, 1174. Townley, R. C. and Gould, I. A. 1943. A quantitative study of heat labile sulfides of milk. I. Method of determination and the influence of temperature and time. II. General origin of sulfides and relation of total sulfur, total nitrogen and albumin nitrogen. III. Influence of pH, added compounds, hydrogenation and sunlight. J. Dairy Sci., 26, 689-843-853. Van der ven, P., Begemann, H. P., and Schogt, J. C. 1963. Precursors of methyl ketones in butter. J. Lipid Res., 4, 91. Van Duin, H. 1958. Investigation into the carbonyl compounds in butter. III. Phosphatide bound aldehydes. Neth. Mulk and Dairy J., 22, 90. Walsh, J. and Merritt, C., Jr. 1964. Qualitative functional group analysis of gas chromatographic effluents. Anal. Chem., 22, 1378. Wheaton, W. and Wentworth, J. 1959. Exhaust gas analysis by gas chromatography combined with infrared analysis. Anal. Chem., 22, 349; Whitney, R. McL. and Tracy, P. H. 1949. Flavor components in dried whole milk. I. The distribution of stale flavor between fractions of reconstituted stale whole milk powder. J. Dairy Sci., 22, 383. Whitney, R. McL. and Tracy, P. H. 1950. Stale flavor- components in dried whole milk. II. The extraction of stale butteroil from stale dried whole milk by organic solvents. J. Dairy Sci., 22, 50. 105 Willitis, C. 0., Ricciuti, C., Ogg, C., Morris, S. G., and Riemenschneider, R. W. 1953. Formation of peroxides in fatty esters. I. Methyl oleate. Application of the polarographic and direct oxygen methods. J. Am. Oil Chem. Soc., 22, 420. Winter, M., Stoll, M., Warnhoff, E., Greuter, F., and Buchi, G. 1963. Volatile carbonyl constituents of dairy butter. J. Food Sci., 26, 554. Wishner, L. A. and Keeney, M. 1963. Carbonyl pattern of sunlight-exposed milk. J. Dairy Sci., 46, 785. Wong, N. P., Patton, S., and Forss, D. A. 1958. Methyl ketones in evaporated milk. J. Dairy Sci., 42, 1699. Wong, N. P. and Patton, S. 1962. Identification of some volatile components related to the flavor of milk and cream. J. Dairy Sci., 46, 724. Wong, N. P. 1963. A comparison of the volatile compounds of fresh and decomposed cream by gas chromatography. J. Dairy Sci., 46, 571. Woods, A.E.anu1Aurand, L. W. 1963. Volatile compounds in ladino clover and off-flavored milk. J. Dairy Sci., 36, 656. wynn, J. D., Brunner, J. R., and TrOUt, G. Mo 1961. Gas chromatography as a means of detecting odors in milk. Am. Milk Rev., 22, 30. HICHIGQN STQTE UNIV. LIBRQRIES Ill" 8 15 HHII 1| 1| ”III “II" 6 312931039 0