FRACTIONATION AND IDENTIHCATQON OF SOME COMPOUNDS IN WOOD SMOKE Thosi: for tho Dam of Ph. D. MEG-{EGAN STATE UNIVERSITY Ray Wayne PM» 1963 This is to certify that the thesis entitled FRACTIONATION AND IDENTIFICATION OF SOME COMPOUNDS IN WOOD SMOKE presented by Roy Wayne Porter has been accepted towards fulfillment of the requirements for Ph.D. Food Science degree in—___ sz//W / Major professé/ Date July 5: 19 63 LIBRARY Michigan State University ABSTRACT FRACTIONATION AND IDENTIFICATION OF SOME COMPOUNDS IN WOOD SMOKE by Roy wayne Porter A smoke generator was constructed with which the generation tempera- ture could be controlled within reasonable limits. The total steam volatile and non-steam volatile phenols, acids, and carbonyls were deter- 'mined on the condensates of whole smoke and the vapor phase generated at 350, 400, 450, 500 and 550°C. The steam volatile monocarboxylic acids generated at 450°C were separated on silicic acid-glycine columns buffered at a given pH. These acids were identified by paper chromatographing their ammonium salts along with the salts of known acids. The steam vola- tile monocarbonyls generated at 550°C were separated as their 2, 4 dini- trophenylhydrazone derivatives on nitromethane-hexane-celite columns. The chain lengths of the parent carbonyls were determined by comparing their column and paper chromatographic behavior with known derivatives. The classes of the parent carbonyl compounds were determined by comparing the absorption maxima of neutral solutions of their 2, 4 dinitrophenylhydra- zone derivatives with the derivatives of known carbonyls. Final identifi- cation was made by observing a lack of depression in the melting points of mixtures of known and unknown derivatives in the instances where the unknown was present in sufficient quantity. The condensates from whole smoke and the vapor phase were analyzed for 3, 4 benZOpyrene by the use of activated alumina columns and the tars from the inside of a commercial smoke generator were analyzed in the same manne r . Roy Wayne Porter A limited study of the comparison of smoke flavor intensity and pre- ference as judged by a taste panel was conducted on cheddar cheese and bacon smoked with whole smoke and the vapor phase. The total phenol production was the highest at a generation tempera- ture of 500°C, total acids were the highest at 450°C, and the peak pro- duction of carbonyls was at 550°C. The Cl to 010 aliphatic monocarboxylic acids were identified in the steam volatile portion of the whole smoke, 'with acetic, formic, prOpionic and butyric acids in decreasing order of occurrence making up the largest portion of the acids. The Cl to C4 acids only were detected by these methods in the vapor phase and their concen- trations ranked in the same order as in the whole smoke. The steam vola- tile monocarbonyls identified were 2-pentanone, valeraldehyde, 2-butanone, butanal, acetone, prOpanal, crotonaldehyde and ethanal, with isovaleral- dehyde and methanal being tentatively identified. No 3, 4 benzopyrene was identified in the condensate from either whole smoke or the vapor phase by the procedure used. A small amount was detected in the tars from a conventional smokehouse generator with the isolated sample showing some contamination as evidenced by its spectrum. The cheese smoked with the vapor phase had a more intense smoke fla- vor than that smoked with whole smoke according to taste panel evaluations, with the Opposite being indicated in the case of the smoked bacon. Yet, no preference was indicated for the cheese or bacon smoked by either the whole smoke or the vapor phase. This indicated that the taste panel did not consider the intensity of smoke flavor to be the major factor in detenmining preference. FRACTIONATION AND IDENTIFICATION OF SOME COMPOUNDS IN WOOD SMOKE By Roy wayne Porter A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Food Science 1963 ACKNOWLEDGMENT The author wishes to exPress his appreciation to Professor L. J. Bratzler for his thoughtful guidance throughout this study and for his aid in preparing the manuscript. He, also, wishes to thank Dr. A. M. Pearson for his advice and encouragement. The author is grateful to Mrs. Beatrice Eichelberger for typing the manuscript and to the Wilbur LaRoe, Jr. Memorial Foundation for financial aid. To his wife, Cynthia, and son, Douglas, the author is especially grateful for their sacrifices and for making it all worthwhile. ii TABLE OF CONTENTS Page INTRODmTION O O C O O O O O O O O O O O O O O O O O O O O O O O O 1 REVIEW OF LITERATURE . . . . . . . . . . . . . . . . . . . . . . . 3 Composition of wood . . . . . . . . . . . . . . . . . . . . . 3 Thermal decomposition of wood . . . . . . . . . . . . . . . . 4 Cellulose . . . . . . . . . . . . . . . . . . . . . . . 5 Hemicellulose . . . . . . . . . . . . . . . . . . . . . 6 Lignin . . . . . . . . . . . . . . . . . . . . . . . . . 6 Smoke generators . . . . . . . . . . . . . . . . . . . . . . 7 Smoldering wood . . . . . . . . . . . . . . . . . . . . 7 Friction generators . . . . . . . . . . . . . . . . . . 8 Smoke generation . . . . . . . . . . . . . . . . . . . . . . 9 Smoke deposition . . . . . . . . . . . . . . . . . . . . . . 11 Modification of the smoking process . . . . . . . . . . . . . 13 Properties acquired during smoking . . . . . . . . . . . . EXPERIMENTALPRDCEDURE............ Smoke generator . . . . . . . . . . . . . Source and treatment of sawdust . . . . . . . . Smoke generation and collection . . . . . . . . . . . . . . . 19 Analysis of smoke condensate . . . . . . Total phenols . . . . . . . . Total acids . . . . . . . Total carbonyls . . . . . . . iii Page Chromatography of volatile acids . . . . . . . . . . . . . . 25 Preparation of silicic acid . . . . . . . . . . . . . . 25 Preparation of the aqueous phase . . . . . . . . . . . . 26 Preparation of the columns . . . . . . . . . . . . . . . 27 Preparation of known acids and Operation of columns . . 27 Preparation of unknown acids . . . . . . . . . . . . . . 29 Column chromatography of unknown acids . . . . . . . . . 30 Identification of unknown acids . . . . . . . . . . . . 31 Column chromatography of unknown monocarbonyls . . . . . . . 35 Identification of the unknown 2, 4 dinitrOphenylhydrazones . 40 Determination of 3, 4 benZOpyrene in smoke condensate . . . . 42 Extraction . . . . . . . . . . . . . . . . . . . . . . . 42 Chromatography . . . . . . . . . . . . . . . . . . . . . 43 Analysis of tar .from conventional smoke generator . . . . . 45 Application of whole smoke and vapor phase to food products . 46 Smoking of cheese . . . . . . . . . . . . . . . . . . . 46 Organoleptic evaluation of the smoked cheese . . . . . . 47 Smoking of bacon . . . . . . . . . . . . . . . . . . . . 48 Organoleptic evaluation of the smoked bacon . . . . . . 50 RESULTS AND DISCUSSION . . . . . . . . . . . . . . . . . . . . . . 51 Effect of generation temperature on total phenols, acids and carbonyls O O O I O I O O O I O O O O O O O O O O O O O O O O 51 Separation and identification of steam volatile acids generated at 450°C 0 O O O O O O O O O O O O O O O O I I O 55 iv Page Separation and identification of steam volatile monocarbonyls from whole smoke generated at 550°C . . . . . . . . . . . . . 67 3, 4 BenZOpyrene analysis . . . . . . . . . . . . . . . . . . 73 Analysis of tar from conventional smoke generator . . . . . . 74 Evaluation of smoked cheese and bacon . . . . . . . . . . . . 77 SUMMARY AND CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . 79 BIBL IOGMPHY . O O O O O O O O O O O O O O O O O O O O O O O O O O 8 1 LIST OF TABLES Table Page 1 Recovery of monocarboxylic acids placed on various pH calms O O O O O O I O O O O O O O O O O O O O O O O I O O 29 2 Smoke treatment of the various bellies . . . . . . . . . . 49 3 Summary of collection data to: samples subjected to analysis for total phenols, total acids and total carbonyls . . . . 51 4 Total phenol concentration in smoke generated at various temperatures 0 O O O O O O O O O O O I O I O O O O O O O O 52 5 Total acid concentration in smoke generated at various twperatures O O O I O I O O O I O O O O O O O O I O O O O 53 6 Total carbonyl concentration in smoke generated at various temperatures 0 O O I O O O I O O O I I O O O O O O O O O O 54 7 Comparison of column chromatographic data for known acids and acids from smoke samples . . . . . . . . . . . . . . . 64 8 Comparison of Rf values for known acids and acids from smOke Sarnples O O O O O O O C O O O O O O O O C O O O O O O 65 9 Quantitative recovery of steam volatile monocarboxylic acids from smoke samples . . . . . . . . . . . . . . . . . . . . 67 10 Column chromatographic data for the 2, 4 dinitrOphenylhydra- zone derivatives . . . . . . . . . . . . . . . . . . . . . 68 11 Paper chromatography of 2, 4 dinitrOphenylhydrazones (heptane- me thanal ) O O C O C C C O O O O O O O . O O O C O O C O C C 7 0 12 Paper chromatography of 2, 4 dinitrOphenylhydrazones (reverse phase, methyl acetate-water) . . . . . . . . . . . . . . . 71 13 Absorption maxima of 2, 4 dinitrophenylhydrazones . . . . . 72 14 Melting point data for 2, 4 dinitrOphenylhydrazones . . . . 73 15 Taste panel evaluation of smoked cheese . . . . . . . . 77 16 Taste panel evaluation of smoked bacon . . . . . . . . . . 78 vi Figure 10 11 LIST OF FIGURES Schematic diagram of Silicic Silicic Silicic Silicic Silicic Silicic Silicic Silicic acid-glycine acid-glycine acid-glycine acid-glycine acid-glycine acid-glycine acid-glycine acid-glycine smoke generating unit . column column column column column column column column pH pH 2.0 (known acids) 8.4 (known acids) 10.0 (known acids) 2.0 (4.0 ml. whole smoke) 8.4 (45 ml. whole smoke) . 10.0 (105 ml. whole smoke) 2.0 (8.0 ml. vapor phase). 8.4 (90 ml. vapor phase) . Absorption spectrum of forerun fraction from activated alumina column . . . . . Absorption Spectrum of isolated hydrocarbon(s) and pure 3, benZOpyrene . . . . vii Page 18 56 57 58 59 60 61 62 63 75 76 INTRODUCTION It is well known that the smoking of food products has a preserving effect on them as well as imparting a desirable color and flavor to them. The preservation is due mainly to the bacteriostatic, antioxidant and drying effect imparted to the food during the smoking process. However, the compounds re3ponsib1e for these attributes are not well known. There is considerable evidence that such factors as generation temperature and availability of air influence the composition of the smoke produced. Also, it has been shown that smoke is an aerosol con- taining condensed particles su3pended in a continuous vapor phase. Evi- dence indicates that the compounds responsible for smoke flavor are imparted by vapor absorption into the surface and interstitial water during the cold smoking of fish. Fish smoked with the vapor phase only do not appear to differ in color, flavor and keeping quality from those smoked with whole smoke. Evidence has shown that the particulate phase of smoke plays a minor role in imparting smoke flavor to foods. Several workers have indicated that the carbonyl, phenol, and acid groups of compounds are important flavor constituents. The steam vola- tile portion of smoke condensate appears to contain most of the character- istic odor and flavor constituents. The phenolic compounds have received the most attention as the constituents most likely responsible for smoke flavor from both a qualitative and quantitative standpoint. However, since the acids and carbonyls appear to be present in smoke in larger quantities than the phenols, a more detailed study of them would seem appr0priate. (1) (2) (3) (4) -2- The objectives of this study were to: Construct a smoke generating unit with which the temperature of smoke generation could be controlled and the particle and vapor phase of the smoke separated electrostatically. Determine if the total amount of carbonyls, phenols, and acids were maintained in the steam distillate of the condensed vapor phase as well as the whole smoke generated at varying temperatures. Study the composition of the steam volatile acids and monocarbonyls by the use of column and paper chromatography. Analyze for the presence of 3, 4 benzoPyrene in both the vapor phase and whole smoke. REVIEW OF LITERATURE Composition of WOod The main components of wood are cellulose, lignin and hemicelluloses and these are located in the cell wall. When considering wood in general, Wise (1952) stated that one can assume it to be roughly one-half cellu- lose, one-fourth hemicellulose and one-fourth lignin. Cellulose is made up of a linear arrangement of D-gluco-pyranose units with B (1-4) glyco- sidic bonds according to Fruton and Simonds (1958). Lignin is defined by Brauns and Brauns (1960) as that constituent of wood when oxidized with nitrobenzene yields vanillin and syringaldehyde. The structure of lignin is unknown and Brauns and Brauns (1960) have presented a discussion of the various prOposed theories. The hemicelluloses of wood are made up of the non-cellulose polysaccharide components of the cell wall accord- ing to Browning (1952). ‘Montgomery ggnal. (1956) have shown that the hemicelluloses of the corn hull contain uronic acid units and give D- glucuronic acid and D-xylose on acid hydrolysis. The principal carbohydrate components of wood are those aforementioned and they are located in the cell wall. However, according to Wise (1952), the water extract of wood contains starch, arabogalactans (yield arabinose and galactose on hydrolysis), pectic materials, and several glycosides. Wise (1952) also stated that aliphatic and aromatic hydrocarbons, terpenes, aliphatic and aromatic acids and their salts, alcohols, phenols, aldehydes, ketones and quinones, esters and ethers may be present in the extraneous components of wood and that certain woods may contain oils, resin acids, sterols, tannins, and cyclic polyhydric alcohols. -3- Hardwoods are used mostly for the generation of smoke for use in smoking foods. Hamilton and Thompson (1959) reported hardwoods to con- tain 40-45% cellulose and 20-30% hemicelluloses. The following data are presented from an analysis of several American hardwoods by Freeman and Peterson (1941) expressed as a percentage of dry wood: Sapwood (%) Heartwood (Z) Holocellulose 76.20 - 80.96 70.85 - 81.58 Cellulose 55.58 - 62.67 50.96 - 64.42 Lignin 12.27 - 20.61 14.82 - 22.26 Ash 0.11 - 0.32 0.21 - 0.84 Soluble in hot water 1.30 - 4.03 0.43 - 2.48 Soluble in alcohol-benzene ~0.97 - 3.31 0.96 - 6.44 where: Holocellulose = all polysaccharides in the wood soluble in hot water - simpler carbohydrates and soluble dyes soluble in alcohol-benzene - waxes and fats Holocellulose less cellulose== hemicellulose Thermal Decomposition of Wood It is necessary to distinguish between the combustion and destruc- tive distillation of wood. Hawley (1952) stated that destructive dis- tillation precedes combustion and is the release of volatile combustible and incombustible products as well as the formation of charcoal when wood is heated in a restricted supply of air. The same author indicated that combustion is the actual burning of the combustible products of destructive distillation (volatile combustibles and charcoal) when their ignition temperature is reached and there is sufficient air available. The smoke used for smoking foods is generally produced, according to the same author, during limited combustion due to minimum air supply, thus allowing a large portion of the volatile combustible products to escape unburned. The terms pyrolysis, thermal decomposition, carbonization, dry dis- tillation, and destructive distillation have all been used quite inter- changeably in the literature, thus, destructive distillation will be used to cover all of these terms. According to Hawley (1923), the actual destructive distillation of wood begins at about 250°C; soon becomes exothermic (around 280°C); and is completed at about 350°C. Fraps (1901), and later, Goos (1952) have summarized a list of over 200 compounds that are formed by the destruc- tive distillation of wood and found in the condensable pyroligneous acid and tars. 'Moisture tends to decrease the speed at which destructive dis- tillation takes place according to Hawley (1923). Later, Hawley (1952) stated that moisture also serves to retard the rate at which combustion can proceed in wood. Cellulose Cellulose, when heated, first breaks down to 1, 6 anhydro-glucose according to Pictet and Sarasin (1918). Irvine and Oldham (1921) have suggested that this is the result of a reaction involving two steps, the first being the hydrolysis of cellulose to glucose and the second the dehydration of glucose to l, 6 anhydroglucose. Pictet and Sarasin (1918) -6- also indicated that further heating causes 1, 6 anhydroglucose to quickly break down into products such as acetic acid, phenols, water and acetone. 'Hemicellulose The hemicellulose from hardwoods is made up mainly of pentosans and the characteristic thermal breakdown products exPected are furan and its derivatives according to Goos (1952). However, Merritt and White (1943) ,obtained a yield of only slightly above 4% furfural from the destructive distillation of oak wood. Goos (1952) indicated that the low yields of furfural are probably attributable to secondary reactions and that under conditions of destructive distillation, the pentosans break down into simpler compounds. Goos (1952) also stated that the pentosans yield larger amounts of acids than cellulose or lignin. Hawley and Wiertelak (1951) and Merritt and White (1943) have shown that the hemicelluloses are the least heat stable of the components of wood and tend to break down first. Lignin In the case of hardwoods, the phenolic compounds are the typical thermal breakdown products formed on the thermal decomposition of lignin according to Goos (1952). The same author indicated that a great many of the phenolic compounds consist of guaiacol and of pyrogallol 1, 3- dime- thyl ether, their homologues and derivatives. Fletcher and Harris (1952) destructively distilled lignin at 400- 445°C for 7.5 hours and obtained 15-257. aqueous distillate from which a heavy tar separated. The aqueous solution contained methanol, acetone '1‘ to ~JaL A \ I; and the following acids: formic, acetic, prOpionic, butyric, valeric, caproic, enanthic, caprylic and benzoic. Steam volatile phenols included phenol, 0 and P cresol, guaiacol, 2, 4-Xy1enol and 4-methy1-, 4-ethyl- and 4-propyl-guaiacol. In addition, there was a quantity of non-steam volatile phenols. Gilbert and Lindsey (1957) found that lignin and cellulose both formed polycyclic hydrocarbons when heated at 650°C for 1 hour in a nitrogen atmosphere. Smoke Generators Smoldering wood Pettit and Lane (1940) developed an experimental kiln for smoking fish wherein the air-flow during the generation of smoke could be con- trolled. Hicks 35 a1. (1941) described a smoking unit which utilized smoldering wood and allowed for live steam to be incorporated into the smoke stream to increase the relative humidity of the smoke and which used high speed fans to move the smoke at high velocities. Nicol (1960) reported the design of a smoke generator utilizing a new principle whereby sawdust was maintained in a state of violent agita- tion in a vertical tube by a continuous blast of hot air. The density, and to a certain degree, the composition of the smoke was regulated by the temperature of the air and the amount of sawdust fed into the stream. For experimental purposes, Spanyar'ggugl. (1960) built a generator consisting of a tube 18 cm. long and 3 cm. in diameter wrapped with a heating tape and insulation. Sawdust was burned in 0.5 gm. charges and the range of combustion temperatures was measured by a potentiometer. Air was metered through the tube during the combustion process. A.simi- 1ar arrangement was used by Husaini and Cooper (1957) in their study. These workers used an iron pipe 36 inches long and 3 inches in diameter packed loosely with sawdust. A stream of air (1-2 p.s.i.) was introduced at one end, the sawdust ignited and allowed to burn from the other end. However, no arrangement was made to measure the actual combustion temp- erature. Simpson (1961) later described a two-stage eXperimental smoke gener- ator designed in Poland whereby blocks of wood were pyrolyzed in a stream of inert gas at any desired temperature up to 450°C and the resulting smoke oxidized in the second stage by mixing with controlled quantities of heated air. Friction Generators Anonymous (1956) described a friction-type smoke generator which utilized a 6 x 6 x 35 inch piece of hardwood standing vertically on a turning disc. The disc was 9 1/2 inches in diameter and consisted of 8 teeth of carbonized steel blades turning at 1750 r.p.m. The amount of smoke produced was controlled by the weights applied to the tap end of the piece of wood. The smoke was sprayed with water and then filtered through a 5 ply metal screen to remove Sparks. Husaini and COOper (1957), using the same principles described above, constructed a small laboratory scale friction smoke generator, which utilized 1.5 x 1.5 x 6 inch blocks of wood, to compare the smoke produced with that from smoldering sawdust. With slight modifications, this type generator was used by weir 25.31, (1961) to determine the suitability of smoke produced by friction generation for use in processing frankfurters. Smoke Generation The smoke used for smoking foods, according to Hawley (1952) is pro- duced during limited combustion in a minimum air supply that allows a large portion of the volatile combustible products to escape unburned. Various workers have shown that generation temperature influences the composition of the resulting smoke and there is general agreement on the Optimum range of generation temperature. Tilgner (1958) recommended a generation temperature below 300°C and Kuriyama (1960) suggested the best range to be 260°C to 310°C. RUSZ (1960) indicated 280°C to 350°C to be the most applicable for smoke generation. Tilgner 35 al. (1960a) used a two-stage smoke generator and suggested a destruction temperature of 400°C and an oxidation temperature of 250°C for the production of a good phenol in the resulting smoke. These work- ers found that the best conditions for acid production was 50°C lower for both the destruction and oxidation stages in smoke generation. Tilgner .g£.§1. (1960b), in another study, found that both the phenol and acid content of the smoke increased in the 300°C to 400°C range. They also found that in the oxidation step the phenol level increased between 125°C and 200°C and then fell below the original level and that the acid content increased through the entire range of oxidation temperatures studied (125°C - 300°C). Spanyar et al. (1960b) found that the products of incomplete combustion remained constant from wood decomposed between 180°C and 440°C. -10- Pettit and Lane (1940) reported that an air flow of 800 cubic feet per hour was required to produce the desired smoke constituents in maximum amounts. These workers found formaldehyde, diacetyl, acetone, furfuralde- hyde, 5~methy1furfuraldehyde, methyl alcohol, ethyl alcohol, phenols, acetic acid and formic acid to be present in smoke condensed in traps. “/1 They also found that under conditions favoring the destructive distilla- tion Of wood, the production of phenols and particularly the production Of aliphatic acids is greatly enhanced. Tilgner 35 al. (1960a) stated that an air surplus factor Of 8 was required for Optimal production of phenolic compounds in smoke. Spanyar 95 al. (1960b) noted that secondary chemical reactions took place in the smoke, notably the polymerization Of oxy- and di- carbonyl compounds. Rusz (1960) studied the composition Of smokes produced by various hard and soft woods and found little difference between them. Spanyar .g§.§l. (1960b) reported no consistent variation in the smokes produced by hard and soft woods. Rusz (1960) and Spanyar 9531. (1960b) also indicated that variations in water content Of the sawdust did not alter the chemical composition of the smoke. Husaini and COOper (1957) found that smoke produced by.a friction generator contained four times as many steam volatile acids, eight times as many steam volatile carbonyl compounds and more steam volatile phenols than smoke from smoldering sawdust. These workers stated that most Of the smoke odor and flavor occurred in the steam volatile fraction Of the smoke condensate, while the non-steam volatile fraction retained most of the color, but negligible smoke odor and flavor. They also found that -11- most Of the acids in the smoke condensate were steam volatile and of these acids, acetic occurred in the highest concentration, followed by formic, propionic and butyric in decreasing amounts. Rusz (1960) reported smoke produced by a friction generator to con- tain four times as many aldehydes, twice as many acids and 15% more phen- 013 than smoke produced by smoldering sawdust. In a similar study, Tilgner gtflgl. (1960c) indicated that dense smoke from a friction genera- tor contained 1.8 times as many carboxylic acids, 1.2 times as many phenols and twice as many carbonyl compounds as smoke from smoldering sawdust. In another study, Tilgner £5 £1. (1960d) found that friction produced smoke contained 1.3 times as many acids and 1.5 times as many carbonyls as smoke from smoldering sawdust. Tilgner (1958b) stated that benZOpyrene and dibenzanthracene are carcinogenic compounds found in wood smoke. He indicated that the forma- tion of these compounds can be minimized by a destruction temperature be- low 300°C or redistilling the smoke condensate under 200°C. Smoke Deposition Foster (1959) reported that smoke consists Of minute visible parti- cles condensed around nuclei which are suspended in a continuous vapor phase. He indicated that the average radius Of these suSpended particles was between 0.1 and 0.2 microns. Tilgner (1958a) stated that the gaseous components Of'WOOd smoke are more important than the colloidal smoke components. Foster and Simp- son (1961) found no significant differences in appearance, flavor and -12.. keeping quality Of normally smoked and vapor smoked kippers. They con- cluded that in the cold smoking process used for fish that the deposition Of smoke flavor appears to be a vapor absorption process. They also found that the absorption Of smoke vapors was enhanced by increased water content in the fish and increased smoke velocity in the smoking chamber. A limited study with bacon by the same workers indicated that the process of smoke absorption was similar to that in fish. Foster gt a1. (1961), in a study to determine the role Of the particle phase in smoke deposi- tion, indicated that it can serve as a reservoir Of compounds, which are released into the vapor phase when the supply of the latter is diminished in the smoke. These authors stated that this contribution amounts to only about 5% of the smoke vapors absorbed. Tucker (1942), using phenol content as an indication Of smoke pene- tration, found most Of the phenol concentrated on the surface Of smoked hams. Linton and French (1945) reported the steam distillable phenols to be a satisfactory measure Of the degree Of smoking, acetone not as accur- ate as phenols and, although considerable quantities Of steam volatile acids are taken up by fish during smoking, their amount was not a good criteria for degree Of smoking. These workers also indicated that the rate Of deposition of smoke on fish increased with increasing smoke velo- city and, although phenol content is not directly related to color devel- Opment during smoking, it does increase with the appearance of the yellow smoke color. Rusz (1959) described a procedure for distilling the phenols and carbonyls from meat tissue and found it to contain 0.5 - 5.0 mg. Z phen- 013, and traces to 1.0 mg. % Of carbonyl compounds. Proctor (1959) has -13- presented an evaluation of various methods Of estimating the compounds deposited during the smoking process. Spanyar and Kevei (1961) conducted a study using 10% gelatin and lard as model absorbents for smoking. They found that gelatin absorbed more acids and carbonyls but less phenol com- pounds than the 1ard and that placing the gelatin in a natural casing considerably reduced the smoke constituents absorbed. Also, they indica- ted that most Of the absorption was on the surface Of the gelatin models and gelatin smoked without a casing and allowed to stand 5 days showed a considerable loss in amount Of smoke constituents originally absorbed, yet during this same period of time there was a slight migration Of some Of the compounds toward the center. These workers found that various natural and synthetic sausage casings possess different characteristic abilities to absorb various groups Of compounds in smoke. 'Modification of the Smoking Process watts and Faulkner (1954) stated that liquid smokes usually are made from a base of a modified form of smoke condensate from hardwood heated in a closed container. Lapshin and Flekchanov (1960) have described the preparation Of a liquid smoke which is used with conventional smoking in the smoke-curing of meats. Hanley‘gtdgl. (1955a) developed a process whereby smoke particles are charged by passing the smoke between Opposed plates with wire grids carrying a high electrical potential (40,000 V). Bacon, maintained at ground potential was susPended between the Opposing plates and the charged smoke particles were deposited on the bacon by electrical attraction. -14- Design and application of the equipment for a continuous smoking system utilizing this principle, with an infrared oven for heating the product, has been reported by Hanley e£_§1, (1955b). The authors reported that a processing shrink of less than 0.5% can be Obtained with this systsm. Foster (1959) has reported the same principle of electrostatic smoking being used in smoking fish, sausages and other meat products. PrOperties acquired During Smoking COOper and Linton (1936) described the Optimum smoking Of fillets to be the point where they possess a smooth, glossy surface, a character- istic smoke flavor, a golden color, and are dried during the process. Hanley §_t_:_ _a_];. (1955a) stated that the functions Of a smokehouse in the 'meat packing industry are smoking, heating and drying. Smoking is intended to improve the color and palatability as well as the keeping quality of the product, according to Tilgner (1957). White 32 21. (1942) reported that the smoking Of bacon reduced the number Of surface bacteria 10,000 times, permitted the fat to resist the formation Of rancidity, and was superior in flavor when compared with un- smoked bacon. These workers concluded that smoked Wiltshire bacon should keep approximately twice as long as unsmoked bacon. In a later study, White (1944) found that smoked Wiltshire bacon could be stored at least 2 months without becoming rancid, whereas unsmoked bacon usually became rancid after 1 month of storage at the same temperature (-1°C to -18°C). Spoilage caused by the formation Of free fatty acids was found to be of little importance. -15- 7 watts and Faulkner (1954) studied the antioxidant effects Of four commercial liquid smokes, and found them to vary from no effect to a pro- nounced antioxidant effect. These workers also found that smoke and ascorbic acid have a synergistic effect as antioxidants. Erdman.g£.al. (1954) used liquid smoke to determine the preservative effect of smoke on fatty fish. They reported much lower peroxide values for fish with added smoke flavor than without. Also, they found that fish, both salted and smoked, had better keeping qualities than fish which had only been salted. The same authors stated that ascorbic acid added to the fish along with salt did not protect the fish against the development Of ran- cidity. However, when small amounts Of smoke were added, the ascorbic acid acted synergistically with the smoke, resulting in a strong antioxi- dant effect. These workers also found that liquid smoke in concentrations of 0.4% and 0.8% proved toxic to pure cultures Of Staphylococcus aureus, Bacillus subtilis and Proteus vulgaris after being exposed to the liquid smoke for various lengths of time. A liquid smoke concentration Of 0.08% decreased bacterial growth but did not step the growth of the three or- ganisms studied. Kemp 2; a1. (1961) found that smoking decreased the develOpment Of rancidity and the formation of free fatty acids in dry-cured hams. These effects increased as smoking temperature increased. Also, they found that increasing temperature of smoking had a tenderizing effect on the hams. Hess (1928) reported a pronounced bacteriostatic action Of smoke on the growth of organisms in fish fillets, but noticed a decided resistance of bacterial spores to the smoke. Jensen (1943) stated that the nonsporing -16- psychrOphiles active during the curing Of bacon are destroyed along with the nonSporing meSOphiles in the course Of the smoking process. However, thermOphiles in sausage items can Spoil the food product if not inhibited by 3.5% sodium chloride, and molds will also grow in many cured and smoked meats, according to the same author. Gibbons ggual. (1954) found that the smoking Of bacon resulted in considerable weight loss and destruction Of bacteria. Smoke density and temperature were the most important factors affecting the bactericidal effect Of the smoke. Hedrick‘ggual. (1960a) in a study to improve the flavor Of smoked cheddar cheese found that a moderate smoke flavor in a medium cured cheese was favored over the same smoke intensity in aged cheese. This final product was best Obtained when 2 lb. bars Of cheese were exPosed to smoke from hardwood sawdust for 4-6 hours at a smokehouse temperature below 85°F. Surface mold occurred after extended storage at temperatures favor- able to mold growth. The addition Of liquid smoke concentrate to whole milk prior to setting did not produce an equally acceptable product. In a later study, Hedrick.g£”§l. (1960b) reported that smoke from hardwood sawdust had a bactericidal effect on the surface organisms Of cheddar cheese. Nest of the destruction of surface bacteria occurred in the first 3 hours of exposure to the smoke. One hour Of eXposure to smoke was effective in destroying viable-yeast and mold on the cheese surfaces. EXPERIMENTAL PROCEDURE Smoke Generator It was necessary to design a smoke generating unit which would pro- duce sufficient smoke to smoke food products for taste panel evaluation, provide a means Of remOving the particle phase Of the smoke, and allow control Of the generation temperature Of the smoke to be analyzed. To accomplish this, an electric hotplate was .employed to heat sawdust which was sifted onto the tOp Of the hotplate by a vibrating pan suspended from a hopper. The amount Of smoke produced was regulated by the rate at which the sawdust was sifted onto the hotplate. TO separate the particle phase from the whole smoke, an electrostatic air filter (Trion 1962) was attached to the enclosed smoke chamber around the hotplate. An on-Off type con- troller utilizing an expanding gas sensing element on the hotplate, wired into the line supplying current to the hotplate gave insufficient control of the generation temperature. Thus, a variable rheostat was wired into the power line to the hotplate and the smoke generation temperature was maintained by manually adjusting the rheostat at a constant rate Of saw- dust flow to the hotplate. The smoke generation temperature was measured by a Leeds and Northrup, null-current potentiometer, calibrated in degrees Fahrenheit with a thermocouple lead going to the surface of the hotplate at the point where the sawdust dropped from the vibrating pan. A schematic diagram Of the smoke generating unit including the elec- trostatic air filter is presented in figure 1. -17- .uso: wawumuoamw Oxoam mo Emuwmwm OHumEOeom .H ounwwm OHH Houuaou umumomnu Houmnnw> > oHH Houoeowuaouon OH manmwum _ m Oumanuom . woos O dsooosuOnH mmmuu OH can Oumunw> tl waauwam unsmsmm Houafim was uwumumOHuOOHm HOQEMSO OxOEm Homaos umsvsum -19- Source and Treatment of Sawdust The sawdust used in this study was from hard maple wood (ag§£;$accharum), Obtained at a local sawmill. The sawdust was screened twice through hard- 'ware cloth having one-quarter inch mesh. It was then Stored in a loosely covered drum and contained 12% moisture before storage. Prior to being used for the production Of smoke the sawdust was heated at 85°C for 12 hours in an Open pan and allowed to cool in the atmOSphere at room temper- ature resulting in a moisture content ranging from 3.50 to 5.18%. Smoke Generation and Collection Before each run the moisture content Of the sawdust was determined by heating previously weighed duplicate samples for 24 hours at 100°C, after cooling in a dessicator the samples were reweighed and the weight loss was calculated as the percent moisture in the original sample. Smoke was generated in the previously described generator at five different generation temperatures, beginning at 350°C and increasing by 50°C intervals up to 550°C. The lower limit Of 350°C was established by the fact that below this temperature, the production of smoke was too in- efficient to have any practical application for the smoking Of food pro- ducts. This conclusion was arrived at following the heating Of sawdust in a 200 m1. boiling flask placed in a mineral Oil bath, with the boiling flask connected to a coiled trap placed in an ethanol-dry ice bath. Heating different samples of sawdust each for a 4 hour interval at temp- eratures starting at 250°C and increasing to 300°C by 10°C Steps did not Show any condensable products until 290°C, and was more pronounced at -20- 300°C. However, complete breakdown did not occur until the bath tempera- ture was raised to 350°C. The upper limit of 550° was the highest temp- erature of generation which could be adequately controlled. The smoke was generated at these temperatures with the sawdust flow as uniform as possible. The Sawdust flow rate was determined before collection Of the smoke was started by collecting the sawdust being Sifted onto the hotplate over two 5 minute intervals, and sawdust flow was expressed as grams per hour. The temperature Of the surface Of the hotplate was observed every 15-20 minutes and only the rheostat setting was varied to maintain the desired temperature. Collection of the smoke was accomplished by four 250 ml. suction flasks connected in series and placed in an ethanol-dry ice bath. The last flask was connected to a vacuum line and the vacuum adjusted so that there was less than 50 ft. per min. air flow at the outlet Of the smoke generator. This resulted in an air flow around the hotplate which was too small to be measured by an air flow meter. The collection period was 12 hours and the extremes Observed in the generation temperature over this period were recorded in each case. Duplicate samples were collected at each generation temperature Of both whole smoke and Of the vapor phase, the latter Obtained by precipitating out the particle phase with the electrostatic air cleaner, which remained Off while collecting whole smoke. Analysis Of Smoke Condensate The condensed smoke was allowed to thaw at 4°C, and the volume of total condensed liquid was measured. Two 5 m1. aliquots Of the condensed -21- smoke were each steam distilled at the rate Of 550 ml. per hour, until 50 ml. Of steam distillate was collected. Steam distillation was done immediately after thawing the smoke condensate. The 50 m1. Of steam dis- tillate contained the Steam volatile portion Of the smoke condensate and was treated as follows: 25 ml. was analyzed for total phenols; 10 ml. was diluted up to 50 ml. with distilled water and analyzed for total acids; 10 ml. was analyzed for total carbonyl compounds; the remaining 5 ml. was discarded. The portion remaining in the distillation flask represented the non-Steam volatile portion Of the smoke condensate and was diluted up to the Same volume Of 250 ml. in each case with distilled water. This was treated as follows: 100 ml. was analyzed for total phenols; 100 ml. was used for the analysis of total acid content; 25 ml. was employed in the analysis for total carbonyl compounds. The remaining 25 ml. was discarded in each case. Total Phenols The steam volatile samples to be analyzed for total phenols were first extracted by a modification of the method of Braus g5 31. (1952). After diluting up to 25 ml. with distilled water, the steam volatile sam- ple was saturated with NaCl and made Slightly acid with 0.1 N. HCl. It was then extracted four times with 25 m1. aliquots of diethyl ether in a separatory funnel. The ether layers were then pooled and subsequently washed with 15 m1. Of a 5% HCl solution and the HCl layer was then dis- carded. At this point the combined ether layers were extracted with four 50 ml. portions Of a 5% NaOH solution and the ether layer was then dis- carded. The NaOH extract was cooled and saturated with Coz by adding dry -22- ice in small pieces until the temperature reached 5°C. This NaOH extract was then extracted four times with 25 ml. aliquots of diethyl ether and the combined ether layers were then dried over anhydrous Na2304. The ether extract was then placed in a 250 m1. boiling flask which was connected tO a water cooled condenser and the ether was distilled Off by placing the boiling flask in an electric heating mantle. The last Of the ether was taken Off very Slowly, and the residue comprised the phenolic compounds which was taken up by thoroughly washing the boiling flask with three 15 ml. aliquots of distilled water. Although some Of the phenolic compounds are more soluble in water than others, it was quite easy to completely dissolve the residue by letting the first 2 portions of water soak in the flask for a few minutes. The non-Steam volatile samples for phenol analysis were extracted in ./ much the same manner except that the solution was not diluted. Also, four 50 ml. aliquots Of diethyl ether were used to extract the solution after V“ it was saturated with NaCl and made acid with HCl. The rest Of the pro- cedure was the same as that used for the Steam volatile samples. The solution containing the phenolic compounds ‘was then analyzed for total V” phenols. The total phenol content of these samples was determined by the method Of Warshowsky gg a1. (1948). The sulfanilic acid reagent used in this procedure was prepared daily as follows: 0.9 gms. of sulfanilic acid was mixed with 9 m1. Of 12 N. HCl and diluted up to 100 ml. with distilled water in a volumetric flask. In another 100 ml. volumetric flask, 5 gms. Of NaN02 was dissolved in distilled water and diluted up to the 100 ml. -23- mark. Three ml. Of the sulfanilic acid solution was then cooled in a 100 m1. volumetric flask in an ice water bath. After cooling, 5 m1. Of the NaNO2 solution was added and the mixture was allowed to cool for another 5 minutes, after which time an additional 12 ml. Of the NaNOz was added. This final mixture was made up to 100 ml. with distilled water and this was designated as the final sulfanilic acid reagent. This solu- tion was kept in a cooler at 4°C and was used after standing at least 15 minutes. The colorimetric determination on the unknown phenol solutions was carried out as follows: aliquots containing 1, 3, 5 and 7 ml. Of the unknown solutions were pipetted into separate test tubes and 1 ml. Of a 20% Naz C03 solution was added to each Sample aliquot. These tubes were then diluted up to 8 ml. by adding distilled water. Two ml. of the cold, final sulfanilic acid reagent was then added to each test tube. The tubes were then Shaken for 2-3 minutes and read against a blank at 425 mu. The blank contained 1 ml. Of 20% Naz C03, 2 m1. Of the cold, final sulfanilic acid reagent, and 7 ml. Of distilled water. The Optical densities of the various unknown solutions were compared with a Standard curve to determine the concentration of phenols in the solution. This standard curve was prepared from phenol solutions varying from 1 to 20 micrograms per ml. The concentration Of total phenols was then calculated as micrograms Of phenol (steam volatile or non-Steam volatile) per ml. Of the original sam- ple of smoke condensate subjected to steam distillation. Total Acids The steam volatile Samples for total acid analysis were placed in -24- 100 m1. beakers and titrated to pH 7.0 with an automatic titrator. The non-steam.volatile samples were placed in 250 m1. beakers for titration. In each case 0.1 N NaOH which was previously Standardized against Stan- dard dilute (0.1 N) H2804 was used to titrate the samples. The concen- tration of total acids was expressed as milliequivalents Of acid (steam volatile and non-steam volatile) per ml. Of the original sample Of smoke condensate which was steam distilled. Total Carbonyls The total carbonyl content was determined by the gravimetric proce- dure Of Iddles and Jackson (1934). The 10 m1. samples from the steam volatile portion of the smoke condensate were each pipetted into large test tubes, followed by the addition Of 10 m1. Of a saturated solution of 2, 4 dinitrophenylhydrazine in 2 N. HCl. This reaction mixture was then allowed to stand for one hour in an icedwater bath. The precipitate was then retained on Whatman # 42 acid washed filter paper which had been previously folded and weighed. The hydrazones were then thoroughly washed with 2 N. HCl and then with water before drying in a dessicator over CaClz and under a slight vacuun. When the hydrazones reached a constant weight, they were weighed on an analytical balance. The nonrsteam.volati1e samples for total carbonyl analysis (25 ml.) were pipetted into 125 erlenmeyer flasks and had 25 m1. Of the 2, 4 dini- trophenylhydrazine solution added to them. Otherwise, the treatment of these samples was identical to that Of the samples from the steam vola- tile portion of the smoke condensate. -25- The relative concentration of total carbonyl compounds was ex- pressed as milligrams Of acetaldehyde (steam volatile and non-steam vola- tile) per m1. Of the original condensed smoke sample subjected to steam distillation. This was done by multiplying the weight Of the 2, 4 dini- trOphenylhydrazone derivative Obtained, by the following ratio: molecular wt. of acetaldehyde (44.05) = .197 molecular wt. Of acetaldehyde 2, 4 dinitrOphenylhydrazone (224.18) Chromatography of Volatile Acids The steam volatile acids produced at a combustion temperature of 450° C were subjected to chromatographic analysis. This particular generation temperature was chosen for chromatographic analysis because it was at this temperature that the production of total acids appeared to be the highest. The method Of Corcoran (1956) was used for the partition chromato- graphy Of the volatile monocarboxylic acids. This method employed the use Of acid-washed, chromatographic grade, 100 mesh Silicic acid as a supporting medium for a Stationary phase Of 2‘M glycine. The glycine had *been previously adjusted with concentrated NaOH or HCl to a selected pH. TO effectively cover the range Of volatile monocarboxylic acids from formic to capric, it was necessary to use three different columns, pH 2.0, pH 8.4 and pH 10.0. Preparation Of Silicic Acid Silicic acid, Mallinckrodt chromatography grade, 100 mesh in 200 gm. lots as needed, was SUSpended in 10 N. HCl for a period Of 36 hours in a -26- 600 m1. beaker covered by a large watchglass. After the SUSpension period, the supernatant liquid was decanted and the silicic acid washed with water to remove the HCl. Approximately 25 to 30 washings with water were nec- essary to make the gel acid free. At this point the silicic acid was washed with absolute methanol until the alcohol washings were neutral to litmus paper and generally required 2 to 3 such washings. The silicic acid was then washed with anhydrous ether and placed in an Open dish in a dessicator over phOSphorous pentoxide. A slight vacuum was maintained in the dessicator, and the Silicic acid was dried for 3 days with several changes of phOSphorous pentoxide. The Silicic acid was then sufficiently dry and was stored in airtight glass jars prior to use. Preparation Of the Aqueous Phase A 2'M stock solution Of the glycine stationary phase was prepared and stored in a refrigerator. The pH Of the stationary phase was very critical and it was necessary to standardize the pH meter before each titration Of aliquots of the glycine phase. Also, it was necessary to use standard buffer solutions to standardize the pH meter which were close to the pH desired in the aqueous glycine phase to be titrated. Be- fore titrating an aliquot Of the 21M Stock solution Of the glycine phase, it was taken from the refrigerator and allowed tO come to room tempera- ture. The glycine phase at pH 2.0 was prepared by adding 1.0 N. HCl to a 25 ml. aliquot Of the stock solution until pH 2.0 was reached. The gly- cine phase at pH 8.4 and pH 10.0 were prepared by adding 1.0 N. NaOH tO -27- 25 ml. aliquots Of stock solution until the desired pH was reached. The glycine phase in each case was titrated just prior to preparation and de- velopment Of the reSpective columns. Preparation of the Columns TO prepare the column, 22 ml. Of the apprOpriate pH glycine phase and 25 gms. Of the acid-washed silicic acid were ground together in a 250 m1. beaker for 5 minutes, using a test tube as a pestle. Seventy-five ml. of the first eluent 1% n-butanol - 99% chloroform, was gradually introduced with constant stirring to this misture to form a smooth Slurry. A Sintered glass disc was placed in the bottom Of a chromatographic column (2.0 cm. x 42 mm.) to retain the Silicic acid. The Slurry was then added in 4 to 5 small increments, with compaction by a wooden cork on the end Of a glass rod between each increment, and the excess solvent is allowed to drain from the end Of the column. Care was taken throughout never to allow the column packing to become dry, and this was accomplished by keeping a slight excess Of eluent on the tap Of the column at all times. At this point, two thicknesses of a paper milk filter disc were cut to fit the inside Of the column and were pressed onto the top Of column to firm and level the tap layer. Preparation Of known Acids and Operation Of Columns Chloroform.solutions of reagent pure monocarboxylic acids were pre- pared and a composite solution of the acids was made, and the concentration of each acid was varied to simplify their identification when eluted from the column. Composite samples Of formic, acetic, prOpionic and butyric -23- acids were prepared to run on the pH 2.0 columns. Samples containing butyric, valeric, caproic and heptylic acids were prepared for the pH 8.4 columns. Composite samples containing heptylic, caprylic, nonylic and capric acids were also prepared to run on the pH 10.0 columns. An aliquot of the composite sample prepared for the reSpective pH column was pipetted onto the tap Of the prOper column and allowed to per- colate into the silicic acid. Simultaneously, the fraction collector (Rinco VE 2002-B24) was started and the collection Of 5 ml. fractions be- gun. The sides of the column were washed three times with 5 ml. aliquots of the first eluent and these washings were each allowed to percolate into the column. The column was then filled with 1% n-butanol 99% chloroform and the rest Of the eluent placed in a 300 m1. separatory funnel connected to the column by a tygon tube. A slight amount Of nitrogen pressure was applied, sufficient to maintain a flow rate of approximately 5 ml. per minute. A total Of 200 ml. Of each Of the three eluents was used, the concentrations Of n-butanol being 1%, 10%, and 25% in chloroform, and was the same for every column. One-hundred and twenty fractions were collected from each column and each fraction was titrated with approximately .03 N NaOH in absolute methanol, using 1% m-cresol purple in 95% methanol as the indicator. A stream of nitrogen was applied through a glass tube drawn out to a fine tip to each test tube for agitation during the titration to exclude the carbon dioxide from the air. The time necessary to complete the deve10p- ment Of one column was approximately 2 hours. TO prevent broadening Of the peaks eluted from the colunns, it was found to be necessary to apply -29- the Sample and develop the column as soon as possible after the column was prepared. The efficiency Of the various pH columns was determined by checking the recovery Of various charges of known acids from these columns. The recovery Of the known acids placed on the columns is given in table 1 along with the particular column on which the acids were resolved. Table 1. Recovery Of monocarboxylic acids placed on various pH columns. pH Of Meq. Of acid Acid column Charge Recovery Difference formic 2. 0 .5550 . 5498 -. 0002 acetic 2.0 .3661 .3654 -.0007 propionic 2.0 .2655 .2666 -+.001l butyric 8.4 .2007 .2018 +.0011 valeric 8.4 .1674 .1688 +.0014 caproic 8.4 .1290 .1305 +.0015 heptylic 10.0 .0650 .0670 +.0020 caprylic 10.0 .0951 .0969 +.0018 nonylic 10.0 .0573 .0592 +.0019 capric 10.0 .0766 .0757 -.0009 Preparation Of unknown Acids The steam volatile monocarboxylic acids were prepared for chromato- graphy directly as their sodium salts by a modification of the method Of Buyske‘ggdgl. (1957). The condensed smoke was extracted immediately upon thawing with four 50 ml. aliquots Of ether. The ether was immediately -30- extracted five times with 100 m1. portions Of 0.5% NaOH and these basic extracts were pooled and the pH adjusted to 8.0 with H2804. TO avoid the addition Of too much H2804, the pH was adjusted down to approximately 8.5 with 24 N. H2804 and then 1 N. H2804 was used to further lower the pH to 8.0. This solution was then concentrated under vacuum using a rotary evaporator, to a volume of less than 1LK>m1. The pH Of the concentrated solution was then adjusted to 2.5 with a saturated solution of tartaric acid. This mixture was then Steam distilled until a volume Of 500 ml. of distillate had been collected. The steam distillate was then titrated to pH 7.0 with 0.3 N NaOH and concentrated tO dryness under vacuum with the use Of a rotary evaporator. Column Chromatography Of unknown Acids The dried sodium salts Of the volatile acids from a given amount Of smoke condensate was taken up in a small amount Of 2 M glycine (generally 4 m1.). An 0.8 ml. aliquot of the glycine phase containing the unknown acids was adjusted to pH 2.0 with 1.0 N HCl. One gram of the previously described, acid-washed silicic acid was then stirred in and the damp pow- der was mixed with 0.5 m1. Of chloroform to produce a jellylike consistency in the mixture. This was then transferred to the tap of a previously pre- pared column (pH 2.0), prepared the same as described earlier except that the paper discs were placed on tOp of the sample and compressed with a cork on the end Of a glass rod. Three 2 m1. portions Of the first eluent were used to wash the sides of the column and each portion was allowed to percolate onto the column before the next was added. The fraction collector was started at the same time the first portion of eluent was placed on the -31- column and 5 ml. fractions were collected. The Operation of the column was the Same as for the known acids, except that .01 N NaOH in absolute methanol was used to titrate the fractions. Since acetic acid was present in by far the greatest quantity in the samples Of smoke condensate, it was necessary to devise a means of Obtaining the longer chain acids in conveniently titratable amounts with- out flooding the column with acetic acid. This was accomplished by first placing the sodiun salts from 15 m1. Of smoke condensate on pH 2.0 colunns as described above, and collecting the first 60 m1. of eluate from the column and neutralizing this with 0.1 N Na0H. By repeating this proce- dure and pooling the neutralized fractions and taking this to dryness under vacuum, it was possible to concentrate the sodium salts Of the acids butyric and higher, since these acids come Off first, leaving prOpionic, acetic and formic behind. Samples Of the sodium salts Of the acids pre- pared in this manner were taken up in 0.8 m1. of 2 M glycine. The pH Of this mixture was adjusted to that of the stationary phase of the column on which it was to be placed (pH 8.4 or 10.0) by the addition of 1.0 N Na0H. The remainder Of the preparation Of these samples and develOpment Of the colunns was identical with that Of the samples run on the pH 2.0 colunns. Elution patterns for the columns were prepared by plotting milliliters Of base required for each 5 ml. fraction collected against the accumulated volume of effluent passing through the column. This was performed on same ples from both whole and the vapor phase of smoke. Identification Of Unknown Acids Identification Of the acids from the smoke samples was based on a -32- comparison Of their retentienvolumes with those Of known acid mixtures when run on separate, but identically prepared columns. This served as tentative identification, and conclusive evidence for the identification Of these acids was gained by chromatographing the unknown acids along with known acids on paper as their ammonium salts. This was accomplished by pooling each elution peak, concentrating under vacuum, making acid with 0.1 N HCl and steam distilling the acid from the indicator. The steam distillate was then made basic by adding excess concentrated NH40Hs and was then concentrated under vacuum to a small convenient volume Of approximately 25 ml. The prepared ammonium Salts Of the unknown acids were first chroma- tographed by the method described by Buyske £5 a1. (1957) and were Spotted along with the ammoniun Salts Of known acids, on a line 4 cm. from the bottom Of a 30 cm. x 43 cm. Sheet of Whatman NO. 1 paper. The Spots were placed 3 cm. apart and the paper was coiled into a cylinder with the edges stapled for rigidity. These Sheets were then allowed to stand in a cylindrical jar, 24 cm. x 46 cm. containing a solvent system of n-butanol- water-prOpylamine (100-15-1, V/V). These chromatograms were developed for 16 hours in an ascending direction using 1500 m1. Of solvent, and the excess primary amine was removed by placing the develOped chromatogram in a 100°C oven for 12 minutes. The acids were then located as their primary amine salts by dipping the paper into a solution of 0.1% ninhy- drin in chloroform giving a purple color on a white background. The Spots were then circled for permanent record, because they faded after a few days. -33- The method Of Kennedy and Barker (1951) involving a solvent system Of 95% ethanol and concentrated NH40H (100-1, V/V) was used in addition. This method allowed the use of the same Samples of ammonium salts that were prepared for the butanol-water-prOpylamine solvent system. Whatman NO. 1 paper was thoroughly washed in a Shallow pan with 1% oxalic acid and rinsed several times with copious amounts Of distilled water and hung by clips to dry at room temperature. Next, 25 x 45 cm. Sheets of this paper were Spotted 2.5 cm. from the bottom and the spots were placed 3 cm. apart. The paper was coiled into a cylinder and stapled for rigidity, then stood in a 24 x 46 cm. jar containing 300 m1. Of solvent and developed in an ascending direction for 7 hours. The develOped chromatograms were then placed in a 100°C oven for 5 minutes. The Spots were located by Spraying the chromatogram with a solution Of 50 mg. Of bromOphenOl blue in 100 ml. of water made acid with 200 mg. of citric acid. The Spots showed up blue on an orange-yellow background and these Spots were circled for permanent record. This solvent system worked well except that it would not resolve nonylic and capric acid mixtures, whereas the butanol- water-prOpylamine system would. All of the paper chromatograms were develOped at room temperature. The Rf values were calculated for both the known and unknown acids run in both solvent systems, Rf being the ratio Of the distance the acid moved to the distance moved by the solvent front. Separation and Identification of Volatile Monocarbonyls The steam volatile monocarbonyls from condensed whole smoke generated at 550°C were converted to their 2, 4 dinitrOphenylhydrazones and subjected -34- tO the column partition chromatographic procedure of Day ggwgl. (1960). This is a modification Of the method Of Kramer and Van Duin (1954) and in- volves the use Of nitromethane as the immobile phase supported on celite, with hexane serving as the mobile phase through the chromatographic column. The carbonyls from.whole smoke generated at 550°C were selected for Study because this temperature more closely resembled commercial smoke generating conditions than lower temperatures. One hundred m1. Of this condensate was steam distilled at atmOSpheric pressure in two 50 m1. por- tions for 4 hours. Each time the distillate was allowed to react with 2 liters Of a solution Of 2 gms. Of 2, 4 dinitrOphenylhydrazine per liter Of 2 N HCl. This reaction mixture was allowed to stand for 12 hours, after which the 2, 4 dinitrOphenylhydrazones were exhaustively extracted from the aqueous mixture with chloroform. The chloroform extract (2 liters) was washed four times with 500 m1. aliquots of distilled water to remove the residual HCl. A dry mixture Of the 2, 4 dinitrOphenylhydra- zones was Obtained by evaporating the chlorofonm under vacuum. This dry ‘mixture was then extracted with hexane equilibrated with 2% Of its volume of nitromethane (nitromethane layer then discarded), the hexane having first been redistilled from KOH pellets (Day ggdgl., 1960) and the frac- tion boiling between 68 and 70°C collected. The equilibrated hexane ex- tract concentrated tO 320 m1. under vacuum, contained the major portion of the monocarbonyl derivatives. The remaining residue was made up of excess reagent and bis -2, 4 dinitrOphenylhydrazones (Day gg al., 1960) and was discarded. -35- Column Chromatography Of Unknown Monocarbonyls The equilibrated hexane extract (320 ml.) was first fractionated into a poorly separated forerun containing the 2, 4 dinitrophenylhydrazones Of the longer chain carbonyls, and individual fractions Of the C1 to C4 car- bonyl derivatives. This was done by chromatographing 20 m1. aliquots Of this extract on 20 gm. columns and required a total of 16 columns. These columns were prepared by the method Of Day g5 El. (1960) by placing 20 gms. Of celite (Johns Manville Analytical Filter Aid) in the mixing bowl of a Waring Blendor, followed by 200 ml. Of equilibrated hexane. The Speed Of the Waring Blendor was regulated by a rheostat until the mixture showed a distinct rolling action, at which time 15 m1. of nitromethane was slowly added to the slurry and the mixing was continued for 10 minutes. After mixing the slurry was placed in a 250 m1. beaker for subsequent ad- dition to the chromatographic column. The column (2.5 x 19.5 cm.) was prepared by first tamping a glass wool plug into the constricted end and aliquots of the Slurry were placed in the column. Between aliquots a slight amount Of air pressure was applied to compact the Slurry. Care was taken always to maintain a solvent layer above the celite to prevent it from becoming dry and to avoid air pockets in the column during packing. When all Of the Slurry had been transferred tO the column, it was packed with air until the solvent layer was about 1 cm. above the celite. A paper milk filter disc was cut to fit the inside diameter of the colunn (2.5 cm.) and the final compaction of the column was accomplished by com- pressing this paper disc onto the tOp of the celite with a cork fastened to the end Of a glass rod. The paper disc served two functions by providing -36- a means Of forming.a firm, level surface Of the celite and allowed the application of the sample and subsequent washing Of the column without disturbing the surface of the celite. The solvent used in packing the column was discarded in each case. The finished column was packed to a height of approximately 15 cm. A slight amount Of pressure was then applied to the colunn to lower the solvent layer level with the surface, at which time a 20 ml. aliquot of the equilibrated hexane extract was placed on the tOp Of the column and gently forced into the surface Of the celite with air pressure. Two 5 m1. aliquots of equilibrated hexane were used to wash the sides of the column with each being forced into the celite with a slight amount of air pressure. The tOp of the column above the celite was then filled with equilibrated hexane, and a separatory funnel (1 liter) was then filled with equilibrated hexane and allowed to drain into the tap Of the column through a tygon tube and the column allowed to develop. The fractions (forerun and individual fractions) were collected visually from the col- umn as it developed, and the correSponding fractions from all 16 columns were pooled. Each Of the individual pooled fractions were concentrated under vacuum to as small a volume as possible without allowing crystalli- zation to occur. These concentrated fractions were then rechromatographed in 15 m1. aliquots on identical (20 gm.) columns as described above and further fractionated into individual bands. The like bands from this second fractionation were then concentrated and applied again to columns (20 gm.) identical to those used in the two previous fractionations. At this point, no further separation occurred, and the volume of mobile phase -37- required to move each band to the lower edge Of the column (when first color is observed in the eluate) was carefully measured for each band. This was used to calculate the threshold volumes for each band for the particular column used. The threshold volume was calculated as the vol- ume of mobile phase (in this case, equilibrated hexane) required per gram Of celite to move the band to the lower edge Of the column. This value was highly reproducible for the fractions Obtained and could be related to the structure of the derivatives (Van Duin, 1957) (Day 25.212: 1960). The individual bands which occurred close to each other on the column in the last two final fractionations (only 2 Of which did) so as to not be adequately separated for visual collection, were collected in 25 m1. fractions Of the eluate and their absorption was scanned in the 325 to 400 mu region with a Beckman DU spectrophotometer. The fractions having maximum absorption at the same wavelength were pooled and were subsequently rechromatographed and by this procedure sufficient quantities Of the pure fractions were Obtained for further study (Day and Lillard, 1960). After separation Of the 2, 4 dinitrophenylhydrazones by column partition chroma- tography, the eluates were dried under vacuum and when possible, the residues were recrystallized from low boiling petroleun ether and the re- sulting crystals or residue was saved for later study. The pooled forerun fraction Obtained from the first column (20 gm.) fractionation was then concentrated under vacuun to a conveniently small volume, without allowing crystallization to occur. This was subjected in 15 m1. aliquots to column partition chromatography designed for C5 and longer derivatives by another type Of column described by Day gg‘al. (1960). -38- These columns were prepared by placing 60 grams Of celite in a Waring Blendor and adding 600 ml. of the redistilled hexane. Mixing was in- creased by adjusting the rheostat until the mixture exhibited a distinct rolling action, then 66 m1. Of nitromethane was slowly added to the slurry and the mixing continued for 10 minutes. This slurry was then packed in a 3.0 x 60 cm. column containing a Sintered glass disc at the bottom to retain the celite. The packing procedure, Sample application, and development was carried out the same as for the 20 gram columns, ex- cept that redistilled hexane was used as the mobile phase. This fraction- ation resulted in a diffuse forerun which was barely recognizable (no attempt was made to identify this fraction), and individually separated bands. The corresponding bands were pooled, concentrated under vacuum and rechromatographed on identically prepared (60 gm.) columns. The volume of mobile phase required in each case, to move a particular band to the lower edge Of the column was noted for the purpose of calculating the threshold volume of each band. The eluates Of the separated bands having the same threshold volumes were pooled, dried under vacuum and, where possible, the residues were recrystallized and Saved for further study. The threshold volumes Of the individual fractions were compared with. literature values for various 2, 4 dinitrOphenylhydrazones on the parti- cular columns used. The 2, 4 dinitrOphenylhydrazone derivatives Of known carbonyl compounds having Similar threshold volunes with the unknown fractions were prepared by the method of Shriner 33 El. (1956). To pre- pare the 2, 4 dinitr0pheny1hydrazine solution, 0.4 gm. of 2, 4 dinitro- phenylhydrazine was placed in a 25 m1. Erlenmeyer flask followed by the -39- addition of 2 ml. Of 12 N H2804. Next, 3 ml. Of water was added drOpwise with swirling until solution was complete, followed by the addition of 10 m1. Of 95% ethanol to this solution. TO prepare the carbonyl solution, 0.5 gm. Of the pure carbonyl compound was dissolved in 20 ml. of 95% ethanol. The preparation Of the 2, 4 dinitrOphenylhydrazone derivative was accomplished by adding the freshly prepared 2, 4 dinitrOphenylhydra- zine tO the carbonyl solution and the mixture was allowed to stand at room temperature. Crystallization Of the hydrazone in each case generally occurred within 10 minutes and the mixture was allowed to stand overnight. The derivative was then retained on acid-washed filter paper and allowed to air dry. Recrystallization of the derivatives was accomplished by placing them in a 125 ml. Erlenmeyer flask with 30 ml. Of 95% ethanol and heating it on a steam bath. If the derivative went into solution immedi- ately, water was Slowly added until the cloud point was reached or, until 5 ml. Of water had been added. In certain instances when the hydrazone did not dissolve, ethyl acetate was added slowly to the hot mixture until the derivative was dissolved. The resulting hot solution was then fil- tered through fluted filter paper and allowed to Stand at room temperature until crystallization occurred (about 12 hours). The melting points Of these known derivatives were determined by using a modified Thiele appar- atus (Robertson and Jacobs, 1962) containing mineral Oil as the bath liquid and a micro Bunsen burner for the heat source. A small amount of the derivative was tapped into the fused end of a 1 mm diameter capillary which was fastened to a thermometer by a rubber band, with care being taken to have the sample next to the thermometer bulb. This was suSpended in -40- the bath and the temperature Of the bath was then raised at about 2°C per minute. The temperature at which the sample started to liquefy and at which it was completely liquid was noted. The derivatives were consi- dered pure when a constant melting point or range was reached which com- pared favorably with literature values. Generally, 2 to 3 recrystalli- zations were sufficient to purify the known derivatives. The known 2, 4 dinitrophenylhydrazones were then chromatographed on 20 or 60 gram celite columns as suggested by Day ggflgl. (1960) and their threshold volumes were determined. These columns were prepared and de- velOped exactly aS previously described for the unknown derivatives. Identification Of the Unknown 2, 4 Dinitrophenylhydrazones Evidence for the identification Of the 2, 4 dinitrOphenylhydrazone derivatives was Obtained by first determining the chain length and the class Of the parent carbonyl compound. Next, the melting points Of the unknown derivatives were Obtained before submitting the small amounts Of the remaining derivatives to mixed melting point determinations. The determination Of the chain length Of each of the parent carbonyl compounds was based on the comparison Of the threshold volumes of the known and unknown derivatives on identically prepared celite columns as previously described. Further evidence for the chain lengths of the car- bonyl compounds was based on the paper chromatographic behavior of the 2, 4 dinitrOphenylhydrazone derivatives. Two procedures were used for the paper chromatography Of the derivatives, the first was the original method Of Huelin (1952) as extended by Schepartz (1961), and the second was the reverse phase procedure of Seligman and Edmonds (1955). -41- Methanol solutions of the unknown 2, 4 dinitrOphenylhydrazone deri- vatives were chromatographed along with the knowns exactly as defined by Schepartz (1961), except that no attempt was made to regulate the temp- erature of the chromatographic chamber. In the procedure of Seligman and Edmonds (1955), the solvent system of methyl acetate-water (10:7) was used in an ascending direction, with Whatman No. 1 paper impregnated with a 5.1% solution of olive Oil (U.S.P. grade) in carbon tetrachloride exactly as suggested by the authors. The absolute Rf values were not very reproducible for both chroma- tographic procedures used, yet the relative movement Of the various derivatives with reSpect to each other was Observed to be quite repro- ducible in most instances. Considerable difficulty was experienced at first by the author in Obtaining clear, well separated Spots especially in the reverse phase procedure. This problem was solved by the applica- tion Of as small a sample as possible which could be observed under an ultraviolet light. The class Of the parent carbonyl compound was determined by Observing the absorption of the 2, 4 dinitrOphenylhydrazone derivatives in the 250 mu to 450 mu range. Chloroform solutions of the unknown as well as the known derivatives were scanned on a Beckman DU SpectrOphotometer. As reported by Jones ggugl. (1956), the absorption peak of the derivative in neutral solution (chloroform), is of considerable value in determin- ing the class Of carbonyl compound involved. The samples Of unknown derivative remaining from the column frac- tionation, where present in sufficient quantity, was subjected to a -42- melting point determination by the identical procedure described earlier. When all Of the data representing chain length (column and paper chroma- tography), class of carbonyl (maximum absorption), and melting point of the 2, 4 dinitrOphenylhydrazone derivative for all Of the unknowns were compared with the same data for the known derivatives, strong evidence was available for identification. However, conclusive identification was Obtained when the mixed melting points of the known and unknown deriva- tives did not Show a depression from the melting point of the known deri- vative. Preparation of the mixture for the mixed melting points was done by placing approximately the same amount of the known and unknown deriva- tives on the concave Side of a watchglass and grinding the sample together by turning another watchglass inside Of the first. This mixture was then tapped into the end of a 1 mm. capillary tube which had been previously fused Off and the melting point determined in a modified Thiele apparatus as described earlier. Determination of 3, 4 BenZOpyrene in Smoke Condensate The method of Latarjet 23 El. (1956) was used to determine the pre- sence of 3, 4 benZOpyrene in the smoke condensate. This method involves the fixation of the hydrocarbon on activated alumina and the eluate checked for its fluorescence in the same region as known 3, 4 benzopyrene. Extraction Whole Smoke generated at 550°C in the smoke generator described ear- lier was condensed in ethanol-dry ice traps as described for the routine analysis of the smoke condensate, except, that the first collection flask contained glass wool to retain most Of the tars. The vapor phase of smoke -43- generated at the same temperature was condensed in the same manner. The smoking period was 12 hours in each case and the traps were weighed before and after the smoke was condensed to determine the amount of condensate. The condensing flasks were thoroughly rinsed with a mixture Of 400 ml. of acetone and 50 ml. Of water and the condensate was completely dissolved. This solution was then extracted with four 150 ml. portions of cyclohexane in a separatory funnel. The cyclohexane extract was in turn washed with two 250 ml. aliquots of water to remove the residual acetone from the cyclohexane extract. The final cyclohexane extract was then dried over CaClz to remove the last traces of water. Chromatography Alumina (Alcoa, grade F-20) was heated at 180°C for 13 hours in an oven to facilitate activation. After activation, the alumina was Stored in a dessicator over phOSphorous pentoxide prior to use. A chromato- graphic column 2.0 x 60 cm was packed to a height of 40 cm with the dry, activated alumina with the column being tapped with a table knife between aliquots to insure compaction. A close fitting paper disc was pressed onto the tOp of the compacted column with a cork on the end Of a glass rod. This served to level and firm the top surface of the column and did not allow the alunina to be distubred when the sample was applied. The cyclohexane extract was then poured onto the column and allowed to percolate through the alumina. Since 3, 4 benZOpyrene has its maximum absorption at 385 mu (Latarjet 25 gl., 1956), the prOportion of the com- pounds absorbing at this wavelength which were fixed on the activated alumina was determined by comparing the absorbance Of the cyclohexane -44- extract at this wavelength to that of the cyclohexane recovered from the end of the column. In each instance 100% Of the compounds which absorb at this wavelength were fixed on the column Of activated alumina. Two 10 ml. portions of cyclohexane were used to wash the inside of the top Of the column and the first was allowed to enter the column packing be- fore the second was applied. After washing, the column was filled with benzene and a separatory funnel filled with benzene was attached to the tOp of the column with tygon tubing. A Slight vacuum was applied at the bottom Of the column by attaching a water aSpirator to a 125 ml. suction flask, which was in turn attached to the column by means Of a cork with the column outlet protruding through it. Two suction flasks were used and each was marked at the 25 ml. level and by alternating the two flasks at the base of the column, 25 ml. fractions were collected. The fluorescence of the fractions was checked with a Fisher Nefluoro- Photometer using a Mercury arc light source. The center filter used was 365 mu, the right filter 450 + mu, and the left filter 425 mu (Operating manual, Fisher Nefluoro-Photometer). The generating solution contained 2.56 x 10"5 gms/ml. of 3, 4 benZOpyrene GNutritional Biochemicals) in benzene and fractions containing 2.30 x 10"8 gms. of 3, 4 benZOpyrene per ml. were set to Show 100% fluorescence, thus, fractions containing 2.30 x 10'10 gms. Of 3, 4 benZOpyrene would Show 1% fluorescence. The extracts of both whole smoke and the vapor phase did not Show any fluorescence in the same region as 3, 4 benZOpyrene at these concen- trations. However, a yellow substance which came off in the first few fractions did not fluoresce. The fractions containing this were pooled -45- (these fractions were chosen visually) and concentrated under vacuum and their absorption scanned in the ultraviolet region on a Beckman DU Spec- trOphotometer in 10% benzene and 90% cyclohexane. Analysis of Tar from Conventional Smoke Generator A sample of tar from the Sides of the outlet of a smoldering sawdust smoke generator was dissolved in the acetone-water mixture and extracted exactly as the smoke condensates. The extract was placed on the 2.0 x 40 cm. activated alunina column described above. It was eluted with benzene and the 25 m1. fractions collected exactly as before. The fractions were checked for fluorescence with the same instrument described earlier. The filter arrangement and light source was the same as described before. The generating solution contained 2.72 x 10'5 ng/ml. of 3, 4 benZOpyrene and the sample set at 100% fluorescence contained 2.72 x 10"9 gms/ml. of 3, 4 benzopyrene. The fractions which Showed fluorescence were pooled and concentrated un- der vacuum using a rotary evaporator with the evaporation flask immersed in 50°C water. The residue was taken up by washing the flask with three 10 m1. portions of cyclohexane. This solution was then placed on a 1.2 cm. diameter column packed to a height of 20 cm. with the dry, activated alumina. This was then eluted with successive mixtures of 5 ml. benzene and 95 ml. cyclohexane, 10 ml. benzene and 90 m1. cyclohexane, 15 ml. benzene and 85 m1. cyclohexane, ----- etc., until no more fluorescence was observed in the 25 m1. fractions collected under vacuum and Observed in the Nefluoro-Photometer as before. The fractions showing fluorescence were pooled and concentrated the same as for the second chromatography, -45- the residue was again taken up in 30 ml. of cyclohexane and poured on a column 0.5 cm. in diameter and packed to a height of 10 cm. with the activated alumina. This was eluted with successive mixtures of benzene and cyclohexane exactly as described above. The fractions Showing fluorescence were pooled and concentrated as before, taken up in 20 ml. of cyclohexane and subjected to two successive chromatographies on columns 2 cm. in diameter packed to a height of 7 cm. with dry, silicic acid (Mallinckrodt chromatography grade, 100 mesh). Elution, collection of fractions and Observation of fluorescence was performed exactly as in the Operation of the last two activated alumina columns except that vacuum was not used. After the fluorescent fractions from the last silicic acid column were pooled and concentrated, the residue was taken up in 10% benzene and 90% cyclohexane. The ultraviolet absorption Spectrum Of this solu- tion was Obtained using a Beckman DU spectrophotometer. Comparison of the ultraviolet absorption Spectrun of the isolated compound with that of pure 3, 4 benZOpyrene in 10% benzene and 90% cyclohexane gave a good indication of the purity of the isolated compound. The concentration of 3, 4 benZOpyrene in the solution containing the isolated compound was determined by comparing the absorbance of this solution against that of solutions containing known concentrations Of 3, 4 benzopyrene. Application of Whole Smoke and Vapor Phase to Food Products Smoking Of Cheese A limited study was undertaken to determine the difference, if any, which existed between the organoleptic quality of aged cheddar cheese smoked with whole smoke when compared with that smoked with the vapor -47- phase. Two 1 1b. blocks Of cheese were suspended on a hardware cloth rack in a laboratory dessicator. A rubber stOpper having a glass intake tube, glass outlet tube and a thermometer protruding through it, was placed in the Opening of the dessicator cover. The glass intake tube was connected to the outlet of the laboratory smoke generator by a tygon tube, and the other end Of the intake tube continued below the level of the hardware cloth and terminated about 1 1/2 inches above the bottom Of the dessicator. The outlet tube, which was connected to a vacuum line, just protruded through the rubber stOpper. Thus, by applying a slight vacuum on the outlet tube, the smoke entered the bottom Of the dessicator and filled the dessicator. This procedure was followed for the smoking of aged cheddar cheese from the Michigan State University Dairy. Two 1 1b. blocks Of cheese were smoked for 4 hours with whole smoke and two 1 lb. blocks were smoked for 4 hours with the vapor phase. The sawdust flow rate remained constant throughout the smoking period and the smoke was generated at 550°C, and the electrostatic air cleaner was turned Off when smoking with whole smoke. The temperature inside the dessicator was 25°C at the start Of the smoking period and reached a maximum of 31.5°C in each instance. After smoking, the cheese was wrapped in paper and stored in a 4°C refrigerator. Organoleptic Evaluation of the Smoked Cheese The smoked cheese was taken from the cooler and the outside 1/16" layer was removed with a knife. Cross sections of the 2" x 2 1/2" bars were then cut approximately 1/8" thick to be subjected to taste panel evaluation. One Slice Of cheese smoked by each Of the two methods was -43- given each Of twelve taste panel judges. Each panel member was given two score sheets, on one sheet the panel member was instructed to rank the two samples on the intensity of smoke flavor present from 1 to 10 (1 least, 10 most); on the other Sheet the panel member was instructed to score the samples as to preference on a hedonic scale (9-like ex- tremely, 5 neither like nor dislike, l-dislike extremely). Smoking of Bacon A Similar approach was undertaken to study the relationship between the organoleptic quality of normally cured bacon, smoked with whole smoke and the vapor phase. The bellies from the right and left Sides of three hogs were Obtained from the Michigan State university Meat Laboratory. These bellies were dry cured in a commercial cure at random with several other bellies. After remaining in cure for 15 days the bellies were removed and soaked for 1 hour in warm (35°C) water. The bellies were then hung by the flank end in a 4°C. cooler until chilled and dried for 6 hours, before the first one was taken out to be smoked. The 6 bellies were put into cure, taken out of cure and soaked all at the same time, however, they were smoked separately. One Side of each pair Of bellies (right or left) was smoked with whole smoke and the other pair member was smoked with the vapor phase. The smoking chamber used was the Meats Laboratory air-conditioned smoke- house and the smoke was provided by connecting the previously described smoke generator to the smokehouse. In order to provide sufficient smoke density, the sawdust flow rate onto the hotplate was too great to facilitate the regulation of the generation temperature. Therefore, the -49- generation temperature was not controlled and only a comparison between the whole smoke and the vapor phase could be made. The collection plates of the electrostatic air cleaner were removed and this unit remained Off when smoking with whole smoke. The collection plates were replaced in the electrostatic air cleaner and the unit turned on when smoking with the vapor phase. Once the sawdust flow rate was set, it was never changed until all the bellies were smoked. The individual belly to be smoked was placed in the smokehouse with a thermometer inserted into the center. The smokehouse was controlled at 136-150°F until the bacon reached an internal temperature of 125°F, at which time the smokehouse temperature was reduced to 130-140°F until a total of 7 hours had lapsed, and the bacon was then removed from the smoke- house and stored at 4°C. Smoke was applied throughout the 7 hour period and smoking Of the Six bellies was completed in a 48 hour period. The treatment of the various bellies is summarized in table 2. Table 2. Smoke treatment of the various bellies. Identification Smoke treatment 15 E left vapor 15 E right whole 05 E left whole 05 E right vapor 16 E left vapor 16 E right whole -50- Organoleptic Evaluation of the Smoked Bacon The bacon Slabs were Sliced anterior to the last rib and 'baked on a rack at 275°F for 50-55 minutes before serving to the taste panel judges. The bacon was presented to the taste panel so that the pairs were compared. The panel members were instructed to evaluate the bacon in the same manner as in the case Of the smoked cheese. Again, twelve taste panel judges were used for evaluation of each pair of smoked bellies and the same panel members were not necessarily used in all three compar- isons. RESULTS AND DISCUSSION Effect Of Generation Temperature on Total Phenols, Acids and Carbonyls The extremes in generation temperatures Observed, sawdust flow rate / and moisture content and volume of condensate collected for analysis for total phenols, acids, and carbonyls are presented in table 3. Table 3. Summary Of collection data for samples subjected to analysis for total phenols, total acids and total carbonyls (average Of duplicate 12 hr. runs). Observed Sawdust Liquid Gen. temp. Sawdust ‘moisture condensate temp. extremes flow-rate content collected (°C) (°C) gms/hr. (7.) (m1.) 3503* 341 - 354 59.5 4.31 22.2 350b** 343.5 - 353.3 63.6 3.96 23.3 400a 394.5 - 403.5 60.8 4.64 21.7 400b 394.5 - 406.5 60.4 3.86 21.5 4503 433 - 451.5 61.1 4.08 23.7 450b 432.4 - 452.6 61.3 4.06 22.5 500a 486.2 - 500 62.6 4.19 21.4 500b 485 - 501.4 59.5 4.07 22.8 5508 537.6 - 551 58.8 4.39 23.6 550b 540.5 - 553.5 62.1 4.43 24.2 a* - vapor phase b** = whole smoke Considerable difference was observed in the appearance of the liquid condensates from the vapor phase and whole smoke. The condensate from whole smoke was extremely dark brown with a somewhat Oily appearance, -51- -52- while that from the vapor phase was much lighter in color. Only minute tar deposits were observed in the cold traps in a few instances after collecting the vapor phase. But considerable deposition of tars was Ob? served in the cold traps after every collection Of whole smoke. This in- dicated that most of the particles were removed by the electrostatic air cleaner during collection of the vapor phase. Also, it must be empha- sized that only the portion Of the condensate which was liquid after the sample thawed was analyzed. Thus, the insoluble tars (present mainly in whole smoke) was not analyzed. However, the liquid portion condensed around the tarry nuclei Of the particles (Foster 1959) was included in the portion Of whole smoke analyzed. This indicates that only the most volatile portion of the smoke was analyzed in the case of the vapor phase. The relative amounts of steam volatile and non-steam volatile phenols in both the vapor phase and whole smoke generated at different tempera- tures are summarized in table 4. Table 4. Total phenol concentration in smoke generated at various temp- eratures. (u gms. phenol/m1.) Gen. Type of smoke temp. Whole Vapor (°C) so VO* No so VO** So V.* N. So VD“ 350 88.0 1.7 56.4 1.2 400 90.8 1.6 69.2 1.3 450 153.0 2.9 138.0 . 1.9 500 196.5 4.7 160.9 1.7 550 165.4 3.5 151.7 1.9 *steam.volatile **non-steam volatile -53- The values Of total phenols were found to be approximately the same for both types Of smoke produced at 350 and 400°C. A significant increase was observed at 450°C and the highest phenol level was obtained at 500°C, with a Slight decrease at 550°C. The generation temperatures at which the highest phenol concentrations were Observed are higher than those sug- gested by Tilgner ggnal. (1960a) and Tilgner SE 31. (1960b). Essentially no differences were found in the prOportion of non-steam volatile phenols from the vapor phase Of smoke over the range of genera- tion temperatures studied. Whole smoke generated at 450°C and above showed a slight increase in non-steam volatile phenols. From the data in table 4 it is Obvious that the largest prOportion of phenols in both the vapor phase and whole smoke are Steam volatile, corresponding with the Obser- vations of Husaini and COOper (1957). The relative amounts Of steam volatile and non-steam volatile acids found in whole smoke and the vapor phase generated at different tempera- tures are presented in table 5. Table 5. Total acid concentration in smoke generated at various tempera- tures_(Meq, acids/m1.) Gen. Type of smoke temp. Whole Vapor (:C) S. V.* N} S. V.** S. V.* N. S. V.** 350 .082 .015 .066 .011 400 .077 .019 .071 .014 450 .121 .023 .115 .019 500 .104 .024 .110 .022 550 .112 .027 .106 .023 *steam volatile **nonrsteam volatile -54- The production of total acids showed the largest value at the 450°C generation temperatures and tended to fall off Slightly at temperatures above this in both the vapor phase and whole smoke. The non-steam vola- tile acids, although a small prOportion of the total acids, increased in amount as the generation temperature increased in both cases. 'The total amount of acids in the whole smoke was slightly higher than in the vapor phase, except for the steam volatile acids generated at 500°C. Thus, it appears that the production of total titratable groups reached a peak at 450°C and above this temperature they possibly started to become oxidized to more stable compounds. Also, the increase in non-steam volatile acids as generation temperature increased would seem to indicate the formation of larger acid compounds at increased generation temperatures. Again, as in the case Of the phenolic compounds, most of the acids were steam vola- tile. The total carbonyl compounds found in the steam volatile and non-steam volatile portions of whole smoke and the vapor phase are presented in table 6. Table 6. Total carbonyl concentration in smoke generated at various temp- eratures. .(mg, acetaldehyde/mlp) Gen. Type Of smoke temp. Whole Vapor (°C) S. V.* N. S. V.** S. V.* N. S. V.** 350 19.95 3.10 12.38 1.46 400 20.18 4.75 19.01 3.43 450 27.29 7.71 25.93 5.07 500 31.17 9.64 27.90 7.77 550 37.18 13.34 34.26 10.11 *steam volatile **non-steam volatile -55- The results in table 6 indicated that there was a distinct tendency for the concentration of total carbonyl compounds to increase as the gen- eration temperature increased for both the whole smoke and vapor phase. This was true for both the steam volatile and non-steam volatile carbonyls. The carbonyl content of both fractions was Slightly higher in the whole smoke than in the vapor phase. This seemed to indicate that at least a part Of the carbonyl compounds are retained in the particle phase. It would be interesting to know if this increase in total carbonyl compounds reached a threshold or peak with increasing generation temperatures above 550°C. Separation and Identification of Steam Volatile Acids generated at 450°C. The steam volatile monocarboxylic acids C1 to C10 were successfully separated on Silicic acid-glycine columns buffered at pH 2.0, 8.4 and 10.0. Typical elution patterns for mixtures Of known acids are presented in figures 2, 3 and 4. AS indicated in these diagrams the longer chained acids are eluted first and the Shorter ones last in each case. On the pH 8.4 column, isovaleric and valeric acids were not separated and butyric and isobutyric acids also were not separated. Therefore, the C4 peaks were labelled butyric-isobutyric and the C5 peaks were labelled valeric- isovaleric. The elution patterns for the steam volatile monocarboxylic acids from various amounts of whole smoke are presented in figures 5, 6 and 7, with the vapor phase being represented in figures 8 and 9. Elution peaks correSponding with all of the C1 to C10 known acids were Obtained from the whole smoke (figures 5-7), whereas, elution peaks corresponding to the C1 to C4 known acids only were Obtained for the vapor phase (figures Ampfiom csocxv o.m me :ESHOO monomawapwom Owowafim .N munwflm seamen .Hz 006 oom ooe com com 00H 4 .. a i - - 3 till. H 7.. 1 O _ e _ . I _ . TL r .f. t. 1: - r. . _ V .et _ _. .m.o " a F. - H _ , m m at K _ u. .5.- D -... v w 3 3 I. #1 .,A J14 D l. m. m- n .m.H 6 a I. T... o O u I. 3 .o.m m m8 5 5 So no Emile—«i m8 :0 Nos '1. zoom smm :osm eon . zoom N- . .m.N 'IW HORN N 87€O° -57- Ammwom czocxv q.w in O ‘0 IO com - owe .| . 1. mac :0 ems zone emm orlfiaanSI-orifinng rliln Ill... aumsnm .-z own IHI m-o 30 Nos zosm so- OTIS {BAOSI-orJSIBA CESHOO mowoxawupwom OHOHHflm OOH - orcadeg .1- rlg mm .d“ 1“ A... I. I: a. A .1 _ ML 1 m - A 1 ; mno :O was :osm ma v .m Onnwwm mo. OH. ma. om. mm. om. mm. on. me. Om. 'IW HOPN N 8980' -58- Ampwom csocxv o.oH In cESHOO ocwozawupwum OHOwHHm .q Ounwwm oumnam .H2 000 com 00¢ oom CON OOH - - .JW q a 4 H 8 v',|||l l J d I .1 . _ LI- 3 rA 1 r I. To 3 D N 3 r I . f e d .L d , 1. rAI- 1. .- K .1 I. _ .l. I. 3 T- 3 . 3 F m J A 1 i mso no ems mso mo Nos m-o mo ems zoom smm . scam KOH Zoom NH in O O .—4 Ln .-.q 0 N mm. om. mm. 0s. 'IN HOBN N 8750' -59- onOEm OHOLB .HE o.qv o.N In cESHOO monoxfiwupflom OHOHme .m Onnmflm mums-m .Hz 6 com ooe com com can 0 ~ 1 a a it. a i m r x. _ f1 J1 _ .H . _.. . l_ .d 1m 0 e 6 a a. _ .m. w . m m Jo.H m. m. w u m n. MW. m.- u. w m:.o.m _ t mom - o.m w . a L m.m 3. I: 3 mno :O ems zoom Nmm zoom NOH zoom NH mno :0 sea .1. Or mso so was a N 10' 'IN HOBN AOxOEm OHOLB .-e mqv q.m ea CESHOO pfiom OHOHHHm .o msnwwm mumnam .H2 000 com :-I ; owe com cam 00- o - J] - - L IIIIIIIIIH m m . OS. .fl n ma r- m. m u. _ . 5 _ a _ 1 m r r s. m m. ., m. r‘ . - o w. fl .fil m. . u n an mom I. 1 D . a ,M a . .w w m. .. A . m. cs. 3 r.- e m fa. A . _ mm _..oe. .f- O - m.__ _ __ 1 ON. _ _ _ - ow. mno no ems zone smm m-o mu mom zonm Nod I; m-o as was zosm u- ll: -_._.V 'IN HOBN N 10' -(11- onoem OHOLB .HE moHv o.oH :n a cESHOO monoxaw u pflom OHOHme .m msswwm moms-m .-z 006 00m 000 00m 00m 00- -| 4 - - 1 om. r__. r—’ J.— 1 L O N r. Fl . -'l “ -e. , . A FA O. CO A B d . . w. u. an 1 .m- 0 .0m. 0 m. a 1. w. n. . q .00. I. 00 AW A .w a . 1 ”we .0e rA .(\ I To 3 I 0w. 4 0a. a . A00 H m-o 00 ems ,_ m-o :0 Nos m-o no Nos .Ilem zone smm .0 :000 so- 000m s- A 'IN HOBN N 10° Aommzn nonm> .HE o.wv o.m mm cESHOO mo«o%aw n nwom OwOHHHm .w madman mums-m .Hz ’5 —(3._ com com com ooN 00H m.o o.H 43 .gI .J °—L1 -J‘“FJF I l ornordoxd Cir—rrfi c: oruordoxd aAoqe J ornaov orwjog L A 0.0 1 A: m-0 :0 was m-o :0 Nos :1 0T mno m0 was. zone emm WT 005m son moss s- O A 'IN H09N N 10' -63- Ammmzm Honm> .HE 00v q.w :n CESHOO ooflomaw u pwum OHOflHHm .m mpnwwm mums-0 .-z 000 00m 00m 00- w. S - a O lllillrllrl :0 A A AOH. M -. O .00 d To m m. A on. W I 8 O m. . . A q. .00 O 1 A .1 TL. 0 A N a .0m. m .6 m Am . A 1 . 00 I. 3 .0k. .00. L :00. 00.- 0-0 :0 ems m-0 :0 s00 mn0 :0 N00 A Ionm HumN zonm KOH - Ion-m ”NH illv -64- 8 and 9). This indicated that the longer chained acids (C5 and above) probably were condensed on the particles removed during the electrostatic filtering process. The acids were tentatively identified by comparing the eluate vol- umes of the peaks obtained from the smoke samples with those of the known acids chromatographed on identically prepared columns. Final identifi- cation was accomplished by paper chromatographing the ammonium salts of the acids from a given elution peak along with the ammoniun salts of the corresponding known acids. A comparison of the eluate volumes of the known and unknown acids from the various silicic acid-glycine columns is presented in table 7, and a summary of the paper chromatographic data is Shown in table 8. Table 7. Comparison of column chromatographic data for known acids and acids from smoke samples. Column Eluate volumg_(ml.) Acid pH Known Whole smoke Vapor phase formic 2.0 290 - 380 275 - 370 305 - 410 acetic 2.0 150 - 260 160 - 260 165 - 285 propionic 2.0 40 - 110 60 - 105 50 - 130 butyric isobutyric 8.4 305 - 380 325 - 420 300 - 430 valeric isovaleric 8.4 200 - 295 195 - 315 - caproic 8.4 55 - 150 60 - 185 - heptylic 10.0 395 - 485 375 - 525 - caprylic 10.0 200 - 310 180 - 310 - nonylic 10.0 80 - 140 50 - 140 - capric 10.0 10 - 70 0 - 40 - -65- Table 8. Comparison of Rf values for known acids and acids from smoke samples Solvent A* Solvent B** Acid Unknown Known Unknown Known formic .31 .32 .31 .31 acetic .36 .34 .35 .36 prOpionic .39 .39 .49 .48 butyric isobutyric .50 .48 .54 .54 valeric isovaleric .57 .58 .64 .62 caproic .67 .66 .68 .69 heptylic .75 .76 .72 .71 caprylic .79 .80 .76 .76 nonylic .82 .82 .82 .81 capric .84 .85 .82 .82 *butanol : H20 : propylamine **95% ethanol : NH4OH (would not resolve nonylic and capric) Again as in the column separation mixtures of butyric and isobutyric acids and of valeric and isovaleric acids could not be separated by the paper chromatographic procedures employed. The ethanol-NH40H solvent sys- tem also was incapable of resolving nonylic and capric acids, whereas the butanol-HZO-propylamine solvent was capable of separating the 2 acids. Table 8 Shows that the mobility of the acids increased directly with their increase in chain length. Considerably larger quantities of samples were required to obtain easily titratable amounts of the longer chained acids separated on the pH 8.4 and 10.0 columns than the shorter acids separated on the pH 2.0 columns. -66- The pH 8.4 columns for the unknown Samples (figures 6 and 9) both indi- cated some propionic acid to be carried over from the initial screening procedure. Husaini and Cooper (1957) reported the presence of fonnic, acetic, propionic, butyric and 2 unidentified acids to be present in both smolder- ing Sawdust and friction generated smoke. These authors reported that all of the total titratable acidity was attributed to these acids. They found acetic acid to be present in the greatest quantity followed by for- mic, prOpionic and butyric in decreasing amounts. The quantitative recovery of the various acids identified in the pre- sent study is summarized in table 9 for the steam volatile acids gener- ated at 450°C. The results Shown in table 9 Show acetic acid to be present in by far the largest concentration, followed by formic, prOpionic, butyric (and isobutyric) acids in decreasing amounts. The acids higher than butyric were detected in extremely small quantities in the whole smoke and were not detected in the vapor phase by the methods used in this study. Formic acid occurred in larger amounts in the vapor phase than in the whole smoke, otherwise, there was little difference observed in the relative prOportions of the C1 to C4 acids in both the vapor phase and whole smoke. In both instances of the whole smoke and vapor phase, all of the titratable acidity in the steam volatile portion of the smoke condensate was not accounted for by the monocarboxylic acids identified (table 9). This would suggest the possibility of the presence of aromatic and even dicarboxylic acids which were not detected by the methods em- ployed. Also, the phenolic compounds present would tend to be weakly acidic in nature, accounting for some of the titratable acidity in the smoke condensate. -67- Table 9. Quantitative recovery of Steam volatile monocarboxylic acids from smoke samples Meq. acid/ml. smoke condensate Acid Whole smoke Vapor phase formic .0193 .0235 acetic .0699 .0700 prOpionic .0065 .0067 butyric isobutyric .0013 .0011 above butyric .0039 --- Total .1009 .1013 Titrated directly .1210 .1150 Separation and Identification of Steam Volatile Monocarbonyls from Whole Smoke Generated at 550°C. The steam volatile monocarbonyl compounds were first separated on nitromethane-hexane-celite columns as their 2, 4 dinitrOphenylhydrazone derivatives. The chain lengths of the parent carbonyl compounds were de- termined by comparing threshold volumes of the unknown derivatives with those of known derivatives on identically prepared columns and by compar- ing paper chromatographic data on the unknown and known derivatives. A summary of the threshold volumes of the various unknown and known deriva- tives together with the type of column used is presented in table 10. The unknown bands were easily collected visually except for the ace- tone and prOpanal bands. This is evidenced by the Similarity in the threshold volumes recorded for these compounds (table 10). As described previously, these 2 bands were collected in 25 ml. fractions and their absorbance scanned in the 325 to 400 mu region. The fractions having the -68- Table 10. Column chromatographic data for the 2, 4 dinitrOphenylhydrazone derivatives Column band 2, 4 dinitrOphenylhydrazone Type Threshold volumep(ml.) No. derivative column Unknown Known Forerun unidentified 60 gm. - - la 2-pentanone 60 gm. 15.5 15.6 b isovaleraldehyde 60 gm. 21.8 21.4 c valeraldehyde 60 gm. 22.5 22.5 2a 2-butanone 20 gm. 23.0 23.4 b butanal 20 gm. 28.7 28.9 3a acetone 20 gm. 38.3 38.1 b prOpanal 20 gm. 38.5 38.3 c crotonaldehyde 20 gm. 43.0 43.4 4a ethanal 20 gm. 51.6 51.2 b methanal 20 gm. 90.3 91.5 most absorbance at 363 mu (acetone) were pooled and those absorbing most at 357 mu.(prOpanal) were also pooled. By subsequent rechromatographing these fractions and repeating this procedure each time, it was possible to separate the 2 bands in quite pure form. The bands were considered pure when only one absorption maxima was present at either 357 or 363 mu. At this point, there appeared to be at least four carbonyl compounds of C5 or longer separated on the 60 gm. columns. This included the fore- run fraction which was barely recognizable and no attempt was made to identify this fraction. The 20 gm. columns resolved the mixture into seven additional carbonyl compounds. The threshold volumes of the unknown -69- derivatives compared well with the known derivatives as indicated in table 10. As chain length of the parent carbonyl compound decreased, the threshold volume of the 2, 4 dinitrOphenylhydrazone derivative increased (table 10). Further evidence for the chain length of the monocarbonyl deriva- tives was obtained by observing their movement on paper chromatograms along with known derivatives. Since the Rf values were not reproducible between runs for either method used, the results of duplicate runs are presented in each case. The results of the paper chromatography of the derivatives utilizing the heptane-methanol solvent system are summarized in table 11. The results in table 11 Show that it was possible to judge only the relative chain length of the unknown derivatives when their Rf values were compared with the known derivatives. AS the chain length of the parent carbonyl increased, the Rf values tended to be higher. The Opposite was true when the derivatives were paper chromatographed using the methyl acetate-water solvent system and the paper impregnated with olive oil (table 12). Again, the Rf values were not sufficient for com- plete identification and were used to aid in confirming the chain length of the parent carbonyls. Jones et a1. (1956) reported that the wavelength of maximum absor- bance of the 2, 4 dinitrOphenylhydrazone derivatives in neutral solutions is a good criterion for establishing the class of the parent carbonyl compound. Aliphatic aldehyde derivatives have their maximum absorbance at 344-358 mu, aliphatic ketones at 364-367 mu, and mono-unsaturated ali- phatic aldehydes have maximum absorbance at 373 mu. Similar values were reported earlier by Roberts and Green (1946). -70- Table 11. Paper chromatography of 2, 4 dinitrophenylhydrazones (heptane- methanol) Column Run 1 Run 2 band 2, 4 dinitrOphenylhydrazone Rf Rf Rf Rf NO. derivative unknown known unknown known forerun unidentified - - - - 1a 2-pentanone .64 .69 .71 .65 b isovaleraldehyde .69 .72 .65 .66 c valeraldehyde .66 .70 .70 .67 2a 2-butanone .56 .55 .55 .60 b butanal .52 .49 .51 .48 3a acetone .43 .40 .46 .44 b prOpansl .38 .40 .42 .47 c crotonaldehyde .42 .39 .43 .40 4a ethanal .30 .25 .32 .28 b methanal .15 .19 .13 .16 -71- Table 12. Paper chromatography of 2, 4 dinitrOphenylhydrazoneS (reverse phase, methyl acetate-water) Column Run 1 Run 2 band 2, 4 dinitrOphenylhydrazone Rf Rf Rf Rf NO. derivative unknown known unknown known forerun unidentified - - - - la 2-pentanone .36 .34 .32 .37 b isovaleraldehyde .28 .31 .32 .30 c valeraldehyde .21 .24 .25 .27 2a 2-butanone .42 .45 .46 .41 b butanal .38 .34 .36 .37 3a acetone .61 .59 .57 .58 b prOpanal .51 .48 .50 .50 c crotonaldehyde .40 .42 .45 .43 4a ethanfll .58 .54 .53 .55 b 'methanal .65 .63 .61 .66 The maximum absorbance of the various 2, 4 dinitrOphenylhydrazone derivatives of the monocarbonyls from the smoke samples are presented in table 13 along with those of known derivatives. These results indicated the presence of one mono-unsaturated carbonyl (band 3c), three aliphatic ketones (bands 1a, 2a and 3a), and six aldehydes (bands 1a, 1b, 2b, 3b, 4a and 4b). Final identification of the carbonyl compounds was accomplished by Observing a lack of depression of the metling points of mixtures contain- ing the unknown and known 2, 4 dinitrOphenylhydrazone derivatives when the unknowns were available in sufficient quantity. The results of the -72- Table 13. Absorption maxima of 2L 4 dinitrOphenylhydrazoneS Column band 2, 4 dinitrOphenylhydrazone Absorption maxima (mp) No. derivative Unknown Known forerun unidentified --- --- la 2-pentanone 362 363 b isovaleraldehyde 358 357 c valeraldehyde 358 358 2a 2-butanone 363 363 b butaral 358 358 3a acetone 363 363 b propaml 357 358 c crotonaldehyde 373 373 4a ethanc al 354 356 b methanal 348 348 melting point data are summarized in table 14 and the melting points are given as the uncorrected melting points. Thus, eight of the eleven ob- served bands from the nitromethane-hexane-celite columns were identified as indicated by the melting point data. Bands 1b and 4b were tentatively identified as isovaleraldehyde and methanal, reSpectively, but were not present in sufficient quantity to allow melting point determinations. The forerun fraction was quite diffuse and was present in too small a quantity to allow further study. However, one would suSpect it to consist of a relatively long chained carbonyl compound(s) Since it was eluted first from the column. -73- Table 14. Melting point data for 2L 4 dinitrophenylhydrazones Column Malting point (°C) band 2, 4 dinitrOphenylhydrazone No. derivative Unknown Known Mixture forerun unidentified - - - la 2-pentanone 141 - 143 142 - 143.5 140.5 -143 b isovaleraldehyde - - - c valeraldehyde 104 - 107 105 - 106.5 103 - 106 2a 2-butanone 112 - 113 113 110 - 113.5 b butanal' 112 - 113 114 - 114.5 112 - 114 3a acetone 124 - 125.5 125.5 123 - 125 b prOpamL 145 - 147 146.5 144 - 146 c crotonaldehyde 186 - 188 188 185 - 188 4a ethanal 164 - 165.5 166.5 163 - 166 b methaml - .. - 3, 4 Benzopyrene Analysis An attempt was made to isolate and detenmine the quantity, if present, of 3, 4 benZOpyrene in the condensates of both the whole smoke and vapor phase generated at 550°C. With the procedure used, no 3, 4 benZOpyrene was found in the condensate of either whole smoke or the vapor phase. The possibility exists that it could have occurred in concentrations too small to be detected by this procedure. Falk and Steiner (1952) stated that polynuclear aromatic hydrocarbons are formed during pyrolysis of many substances in a restricted supply of air in the temperature range of 750°C to l600°C. Thus, a combustion temperature of 550°C was possibly not suf- ficiently high to bring about the formation of 3, 4 benzoPyrene. -74- A yellow substance was observed in the first four to five 25 m1. fractions eluted with benzene from the 40 cm. activated alumina. These fractions were chosen visually and a spectrum was Obtained in the ultra- ‘violet region which appears in figure 10. The absorption peak at 275 mu suggests the presence of an aromatic substance, as several aromatic com- pounds tend to absorb strongly in this region, particularly some aromatic hydrocarbons and quinone type compounds (Friedel and Orchin 1951). The author observed that the amount of this substance found in the whole smoke was considerably greater than that which occurred in the vapor phase. This could reflect one of the largest differences between the whole smoke and the vapor phase. Analysis of Tar from Conventional Smoke Generator The spectra of pure 3, 4 benZOpyrene (2.75 ug/ml.) along with that of the hydrocarbons isolated from the tars of a conventional smoke gener- ator are presented in figure 11. The tars were found to contain 23.1 u gms per gm of tar of the isolated hydrocrabon(s), using pure 3, 4 benzo- pyrene as a Standard. As indicated by the Spectra in figure 11, the isolated fraction appears to consist mainly of 3, 4 benzopyrene, but appears to be contaminated possibly with anthracene, which has peaks at 372, 375.8, and 380 mu (Commins 1958). Distinct Shoulders appear on the curve in the region of these wavelengths and this as well as several other hydrocarbons are known to be eluted with mixtures containing 3, 4 benZOpyrene from ac- tivated alumina columns (Commins 1958). Thus, it was not possible to isolate 3, 4 benZOpyrene from smoke tars in pure form by this method. If one could check the fluorescent Spectrum of each fraction eluted from the Absorbance 1.1 1.00} .90 F .70» / .60 . .50 » .40 » .30 V .20 r .10- 0 .A_______- A ___, 1 . . . A 1 _ . 220 240 260 280 300 320 Wavelength (mu) Figure 10. Absorption smactrum of forerun fraction from activated alumina column Absorbance -76- ‘—‘———‘ Pure 3, 4 benzopyrene X—-—-mx Isolated hydrocarbon(s) from tars .66. /\ V‘ny '. 350 360 370 380 390 400 Wavelength (mu) Figure 11. Absorption Spectrum of isolated hydrocarbon(s) and pure 3 . 4 be nz (my re no. . -77- column and collect only those whose fluorescent Spectra compared with 3, 4 benzopyrene, the fractions containing compounds which have fluores- cent characteristics Similar, but not identical, to those of 3, 4 benzo- pyrene could possibly be eliminated. Evaluation of Smoked Cheese and Bacon The results of the taste panel evaluation of cheese smoked by both whole smoke and the vapor phase are summarized in table 15. Table 15. Taste panel evaluation of smoked cheese Smoking Smoke method intensity Preference whole smoke 3.8 7.3 vapor phase 8.9 7.1 These results indicated that the vapor phase tended to penetrate the cheese more than whole smoke Since a much greater smoke intensity was ob- served in the cheese smoked with the vapors only. One must keep in mind that the smoking of cheese is a "cold" smoking process (i.e. 32°C) and the more volatile constituents would tend to penetrate more readily under these conditions. A somewhat Sticky coating was observed to be more pre- valent on the cheese smoked with whole smoke than that smoked with the vapor phase. This might have impaired the penetration of vapors into the cheese during smoking with whole smoke. The results in table 15 indicated that level of smoke intensity did not have a great deal of influence on the preference of the cheese, Since essentially no difference was found in the preference of the cheese smoked by the two methods. -78- The results of the taste panel study involving the smoked bacon are presented in table 16. Table 16. Taste panel evaluation of smoked bacon Whole Vapor Intensity Preference Intensity Preference 5.4 7.8 3.6 7.1 6.1 7.6 3.8 6.7 5.8 9...; L8. 1.9 Average 5.8 7.2 4.1 7.1 There was a reversal of the situation observed in the smoked cheese in that the bacon smoked by whole smoke was judged to have a consistently higher smoke intensity as indicated in table 16. This could be possibly due to the higher temperatures used in the smoking chambers and longer smoking time for the bacon than for the cheese. In two of the three cases observed, the bacon smoked with whole smoke was preferred to that smoked by the vapor phase whereas, in the third instance the Opposite was true. However, the averages for the preferences indicated for the smoked bacon were essentially the same for both methods (table 16). De- finite conclusions cannot be drawn as to the advantages of the two methods on the basis of this rather limited study of acceptance. SUMMARY AND CONCLUSIONS ‘Most of the phenols, acids and carbonyls present in the smoke con- densates were steam volatile. The highest total phenol level was attained at a generation temperature of 500°C, highest acid level at 450°C and highest level of total carbonyls at 550°C. Column and paper chromato- graphy revealed the C1 to C10 aliphatic monocarboxylic acids to be pre- sent in whole smoke with acetic, formic, propionic and butyric in decreas- ing order of concentration, making up most of the acids. The acids longer than C4 tended to be present in decreasing amounts as chain length in- creased and made up a relatively small proportion of the total acids. Using the same procedures, acetic, formic, propionic and butyric in de- creasing amounts were found to be present in the vapor phase. These were the only acids detected in the vapor phase by the methods employed. In either the case of whole smoke or the vapor phase, all of the titratable acidity in the steam volatile portion of the smoke condensate was not accounted for by the acids identified. The following monocarbonyls were identified in the steam volatile portion of whole smoke: 2-pentanone, valeraldehyde, 2-butanone, butanal, acetone, prOpanal, crotonaldehyde and ethanal.. Isovaleraldehyde and methanal were tentatively identified. No 3, 4 benZOpyrene was detected in the condensates from either the whole smoke or the vapor phase with the procedure used. A small amount (23.1 u gms/gm.) of 3, 4 benZOpyrene was detected in the tars from a con- ventional smoke generator, although the Spectrun of the sample isolated showed considerable contamination. -79- -80- Cheese smoked with the vapor phase was judged to have a more intense smoke flavor than that smoked with whole smoke by a taste panel, whereas, the opposite was found for smoked bacon. However, there was no prefer- ence in either case for that smoked by either the whole smoke or vapor phase indicating that intensity of smoke flavor is not the major factor in preference. BIBLIOGRAPHY Anonymous. 1956. Fireless smokehouse smoker - Its' advantages - how it works. Food Eng. .28, 65. Brauns, F. E., and D. A. Brauns. 1960. The Chemistpy'2§_Lignin. Suppl. Vol., 7. Academic Press, New York. Braus, H., F. M. Middleton, and C. C. Ruchhoft. 1952. Systematic analy- sis of organic industrial wastes. Anal. Chem. 34, 1872. Browning, B. L. 1952. 'The polysaccharide fraction of wood. 2. Analysis of non-cellulose polysaccharides. In Wood Chemistry. (L. E. Wise and E. C. Jahn, eds.) Vol. II, 1159. Reinhold Publishing Corp., New York. Buyske, D. A., P. Wilder, Jr., and M. E. Hobbs. 1957. Volatile organic acids of tobacco smoke. Anal. Chem. .22, 105. Commins, B. T. 1958. A modified method for the determination of poly- cyclic hydrocarbons. Analyst. .83, 386. COOper, D. Le B., and E. P. Linton. 1936. The preparation of fresh fillets of fish for smoking. J. Biol. Bd. 3, 1. Corcoran, G. B. 1956. Chromatographic separation and determination of straight chain saturated monocarboxylic acids C1 through C10 and dicarboxylic acids C11 through C16. Anal. Chem..2§, 168. Day, E. A., R. Bassette, and M. Keeney. 1960. Identification of volatile carbonyl compounds from cheddar cheese. J. Dairy Sci. 43, 463. Day, E. A., and D. A. Lillard. 1960. Autoxidation of milk lipids. I - Identification of volatile monocarbonyl compounds from autoxidized milk fat. J. Dairy Sci. 43, 585. Erdman, A. M., B. M. Watts, and L. C. Ellias. 1954. Smoke flavor and ascorbic acid as preservatives for fatty fish. Food Tech..§, 320. Falk, H. L., and P. E. Steiner. 1952. The identification of aromatic polycyclic hydrocarbons in carbon blacks. Cancer Res. 12, 30. Fisher, Three-way analysis with the Fisher Nefluoro-Photometer. Fisher Sci. 00., New York. Fletcher, T. L., and E. E. Harris. 1952. Products from the destructive distillation of Douglas-Fir lignin. TAPPI. .32, 536. Foster, W. W. 1959a. Attenuation of light by wood smoke. Brit. J. Applied Physics. '19, 416. -81- -82- Foster, W. W. 1959b. Technology of smoked foods. Food Manuf. 34, 56. Foster, W. W., and T. H. Simpson. 1961. Studies of the smoking process for foods. I - The importance of vapors. J. Sci. Food Agric. 12, 363. Foster, W. W., T. H. Simpson, and D. Campbell. 1961. Studies of the smoking process for foods. II - The role of smoke particles. J. Sci. Food Agric..1g, 635. Fraps, G. S. 1901. The composition of a wood oil. J. Am. Chem. Soc. 22, 26. Freeman, R. D., and F. C. Peterson. 1941. Proximate analysis of the heartwood and sapwood of some American hardwoods. Ind. Eng. Chem., Anal. Ed. 13, 803. Friedel, R. A., and M. Orchin. 1951. Ultraviolet Spgctra 9f_Aromatic Compounds. John Wiley and Sons, New York. Fruton, J. S., and S. Simmonds. 1958. General Biochemistry. 2nd ed. 422. John Wiley and Sons. Gibbons, N. E., D. Rose, and J. W. Hopkins. 1954. Bactericidal and drying effects of smoking on bacon. Food Tech. '8, 155. Gilbert, J. A. S., and A. J. Lindsey. 1957. The thermal decomposition of some tobacco constituents. Brit. J. Cancer..11, 398. Goss, A. W. 1952. The thermal decomposition of wood. In WOod Chemistry. (L. E. Wise and E. C. Jahn, eds.) Vol. II, 826. Reinhold Publishing Corp., New York. Hamilton, J. K., and N. S. Thompson. 1959. A comparison of the carbo- hydrates of hardwoods and softwoods. TAPPI‘42, 752. Hanley, J. W., H. N. Draudt, and M. C. Brockman. 1955a. A continuous process for smoked meat. I - Development of a process. Food Tech. 9 591. _) Hanley, J. W., G. L. Montgomery, M. S. Rarick, and M. C. Brockman. 1955b. A continuous process for smoked meats. II - Equipment design and application. Food Tech. 2, 597. Hawley, L. F. 1923. WOod Distillation. Chemical Catalog Company, Inc., New York. 64. Hawley, L. F. 1952. Combustion of wood. In WOod Chemistry. (L. E. Wise and E. C. Jahn, eds.) Vol. II, 817. Reinhold Publishing Corp., New York. -83- Hawley, L. F., and J. Wiertelak. 1931. Effect of mild heat treatment on the chemical composition of wood. Ind. Eng. Chem. .23, 184. Hedrick, T. I., L. J. Bratzler, and G. M. Trout. 1960a. Smoke-flavored cheddar cheese. IMich. Agr. Expt. Sta. Quart. Bull..42, (4) 714. Hedrick, T. I., L. J. Bratzler, and G. M. Trout. 1960b. Lethal effect of smoke on surface microorganisms of cheddar cheese. ‘Mich. Agr. Exp. Sta. Quart. Bull. 43, (2) 327. Hess, E. 1928. The bacterial action of smoke. (as used in the smoke- curing of fish). J. Bact..1§, 33. Hicks, E. W., M. C. Taylor, and E. J. Ferguson Wood. 1941. Note on experimental kiln for the smoke curing of fish. J. Austr. Council Sci. Ind. Res. 14, 308. Huelin, F. E. 1952. Volatile products of apples. III. Identification of aldehydes and ketones. Austr. J. Sci. Res. B2. 328. Husaini, S. A., and G. E. Cooper. 1957. Fractionation of wood smoke and the comparison of chemical composition of sawdust and friction smokes. Food Tech. 1;, 499. Iddles, H. A., and C. E. Jackson. 1934. Determination of carbonyl com- pounds by means of 2, 4 dinitrOphenylhydrazine. Ind. Eng. Chem., Anal. Ed. 9, 454. Irvine, J. C., and J. W. H. Oldham. 1921. The constitution of polysacc- harides. Part III. The relationship of l-glucosan to d-glucose and to cellulose. J. Chem. Soc. 119, 1744. Jensen, L. B. 1943. Action of hardwood smoke on bacteria in cured meats. Food Res..§, 377. Jones, L. A., J. C. Holmes, and R. B. Seligman. 1956. SpectrOphotometric studies of some 2, 4 dinitrophenylhydrazones. Anal. Chem. .28, 191. Kemp, J. D., W. G. Moody, and J. L. Goodlett. 1961. The effects of smoking and smoking temperatures on the shrinkage, rancidity deve1- Opment, keeping quality, and palatability of dry-cured hams. Food Tech. 12, 267. Kennedy, E. P., and H. A. Barker. 1951. Paper chromatography of vola- tile acids. Anal. Chem. lgg, 1033. Kramer, P. J. C., and H. Van Duin. 1954. The separation and identifica- tion of normal aliphatic aldehydes and methyl ketones. Rec. Trav. Chim. _7_3, 63. -84- Kuriyama, A. 1960. The thermal decomposition of woody substances. In A.M.I.F. Special Report No. 26. Lapshin, I. I., and G. V. Flekchanov. 1960. New technology of cold smoking of fish using smoke solution. In A;M.I.F. Special Report No. 26, 29. Latarjet, R., J. L. Cusin, M; Hubert-Habart, B. Muel and R. Royer. 1956. Detection quantitative du 3, 4 benZOpyrene forme par combustion du. papier a cigarettes et du tabac. Bull. De L'Assoc. Fran. Pour L'R'tude Du Cancer 43, 180. Linton, E. P., and H. V. French. 1945. Factors affecting deposition of smoke constituents on fish. J. Fish. Res. Bd. Can. _6, 338. Merritts, R. W., and A. A. White. 1943. Partial Pyrolysis of Wood. Ind. Eng. Chem. 32, 297. ‘Montgomery, R., F. Smith, and H. C. Srivastava. 1956. Structure of corn hull hemicellulose. I. Partial hydrolysis and identification of 2-0-( -D-g1uc0pyranosyluronic acid)-D-xylopyranose. J. Am. Chem. Soc. .18, 2837. Nicol, D. L. 1960. New type of smoke generator. Food Manuf. .22, 417. Pettit, A. E., and F. G. Lane. 1940. The chemical composition of wood smoke. J. Soc. Chem. Ind. 52, 114. Pictet, A., and J. Sarasin. 1918. Sur la distillation de la cellulose et de 1'amidon sous pression raduite. Helv. Chim. Acta. 1, 87. Proctor, B. E. 1959. Chemical and physical procedures for defining quantitatively the degree of smoking of meat products. In A.M.I.F. Special Report No. 26, 18. Roberts, J. D. and C. Green. 1946. Absorption Spectra of some 2, 4 dinitrOphenylhydrazones. J. Am. Chem. Sco. ‘68, 214. Rusz, J. 1959. Determination of phenols and aldehydes in smoked meats. Food Manuf. .34, 59. Rusz, J. 1960. Investigation of suitable conditions in the production of smoke for the smoking of meat products. A.M.I.F. Special Report No. 26. Schepartz, A. I. 1961. Paper chromatography of 2, 4 dinitrOphenylhydra- zones. Extension of the Huelin Method. J. Chromatog. .6, 185. Seligman, R. B., and M. D. Edmonds. 1955. The separation of 2, 4 dini- trophenylhydrazones by paper chromatography. Chem. and Ind. 1406. -85- Shriner, R. L., R. C. Fuson, and D. Y. Curtin. 1956. The Systematic Identification of Organic Compounds. John Wiley and Sons, New York. 219. Simpson, T. H. 1961. Smoke curing technologists meet in Poland. Food Manuf. ‘36, 115. Spanyar, P., and E. Kevei. 1961. Uber Fragen des Raucherns von Leben- smitteln III - Machanismus des Raucherns. Z. Lebens. - Untersuch. und - Forsch. 115, 1. Spanyar, P., E. Kevei, and M. Kiszel. 1960a. Uber Fragen des Raucherns von Lebensmitteln. I - Bestimmung von Rauchbestandteilen und Entwicklung einer Rauchversuchseinrichtung. Z. Lebens. - Untersuch. und - Forsch. 112, 353. Spanyar, P., E. Kevei, and M. Kiszel. 1960b. Uber Fragen des Raucherns von Lebensmitteln. II - Zusamensetzung des Ranches und dessen Beein- flussung durch Faktoren der Rauchbildung. Z. Lebens. - Untersuch. und - Forsch. .112, 471. Tilgner, D. J. 1957. A rational procedure for the hotsmoke curing of fish. Food Manuf. .32, 365. Tilgner, D. J. 1958a. Obtention et utilisation de la fumae destinee au fumage de produits alimentaires. Die Fleischwirtschaft. .19, 758. Tilgner, D. J. 1958b. Smoke curing processes. Die Fleischwirtschaft. 1:9, 649. Tilgner, D. J., A. Makowiecki, and W. Barchowiec. 1960c. A comparison of curing smoke quality as influenced by methods of smoke production. In.A.M.I.F. Special Report No. 26, 7. Tilgner, D. J., K. Miler, J. Prominski, and G. Darnowska. 1960a. The sensoric quality of phenolic and acid fractions in curing smokes. In A.M.I.F. Special Report No. 26, 6. Tilgner, D. J., K. Miler, J. Prominski, and G. Darnowska. 1960b. The influence of smoke production parameters upon the quantity of phenols and acids. In A.M.I.F. Special Report No. 26, 8. Tilgner, D. J., H. Ziminska, and H. Wronska - Wojciechowska. 1960. The chemical and sensorial characteristics of the steam.volatile compon- ents in sawdust and friction smoke. In A.M.I.F. Special Report No. 26, 24. Trion, 1962. Trion electronic air cleaner - model 24-102. -86- Van Duin, H. 1957. Chromatography by liquid-liquid partition and liquid- 1iquid interface absorption. Nature 180, 1473. Warshowsky, B., E. J. Schantz, and'M. W. Rice. 1948. Determination of phenol and m-cresol in complex biochemical mixtures. Anal. Chem. 22, 951. Watts, B. M., and'M. Faulkner. 1954. Antioxidant effect of liquid smokes. Food Tech. 8, 158. Weir, C. E., D. M. Doty, L. J. Pircon, and G. D. Wilson. 1961. A fric- tion-type smoke generator, performance and application for smoking frankfurters. A.M.I.F. Bull. 41, 16 pp. White, WL H. 1944. Smoked meats. II - Development of rancidity in smoked and unsmoked Wiltshire bacon during storage. Can. J. Res. 22F, 97. White, W. H., N. E. Gibbons, A. H. WOodcock, and W. H. Cook. 1942. Smoked meats. I. Bacteriological, chemical, and physical measure- ments on smoked and unsmoked bacon. Can. J. Res. 20D, 263. Wise, L. E. 1952. Extraneous components of wood. In Wood Chemistry. (L. E. Wise and E. C. Jahn, eds.) Vol. I, 543. Reinhold Publishing Corp., New York. um USE cm milllrll' muhwmu MN] U"o Ems Amg 5”” 3 J WW“