Detection and Evaluation of Fragrances by Human Reactions Using a Chemical Sensor Based on Adsorbate Detection Kenji Yokoyama. and Fumihiro Ebisawa Photonic Materials Laboratory, N T T Opto-electronics Laboratory, Tokai, Ibaraki 319-11, Japan
Hman reactknrand sensor responsw to 37 typkaifragrances are InvestigatedIn order to develop a chemkal sensor whlch can detect fragrances and evaluate them in subjectlve t e r n of human reactions. EbMfemale test subjects recordedtheir reactions to these fragrances by grading them on a sevenpoint scale In terms of 5b descriptive characterktlcs. Their reactlonr fell Into two generalcategories, “pleasantness”and “freshness (degance)”, dependent on the characterktk chemkal structure of the fragrances. The women evaluate the fragrances malnly by the degree of pleasantness and frerhness (elegance). A set of elght piezoelectrlc sensors lncorporatlng different hydrophobic polymer materials ls fabrlcated to predict such human reactions h a w r e their adsorbate detectlon and pattern recognition are similar to those In humans. The sensors were effectlve in detecting hydrophoMcfragrances in the gas phase, and their fragrance ciarrlflcatlons were similar to human ones. The sensor responsesto the adsorbate correlate rlgnlflcantlywith two of the human responses with multiple correlation coefficlents of more than 0.6 ( p < 0.02). Namely, the sensor can predlct two different human reactions to fragrances rlmuitanwudy, suggesting th. posslbUlty of devdoplng an artlfkial olfactory sensor basod on adsorbate detection.
INTRODUCTION Recently, great interest has been shown in chemical sensors which are currently in use for detecting toxic gas. A possible future development is an ‘artificial nose” which can perceive odorsin a similar way to a human noee.1,2 The use of chemical sensor arrays in conjunction with the pattern recognition method is a promising approach for developing such sensing devices because the pattern processing technique enables the sensor to detect odor molecules and their chemical properties qualitatively.3-10 The sensing device, which we call an “artificial olfactory sensor”, is expected not only to detect high molecular weight compoundssuch as fragrancesbut also to evaluate them in subjective terms of human reactions. To the authors’ knowledge no sensor has been reported to date. (1) Newman, A. R. Anal. Chem. 1991,63,585A-5881\. (2) Okahata, Y.; Shimizu, 0. Langmuir 1987,3, 1171-1172. (3) Carey,W. P.;Beebe,K. R.; Kowalski,B. R.; Illman,D. L.; Hirschfeld, T. Anal. Chem. 1986,58,149-153. (4) Ballantine, D. S., Jr.; Rose, S. L.; Grate, J. W.; Wohltjen, H. Anal. Chem. 1986,58, 3058-3066. (5) Rose-Pehrsson, S. L.; Grate, J. W.; Ballantine, D. S., Jr.; Jure, P. C. Anal. Chem. 1988,60,2801-2811. (6) Ballantine, D. S., Jr.; Wohltjen, H. Anal. Chem. 1989, 61, 704A-
”. IIZA. I
(7) Grate, J. W.; Klusty, M. Anal. Chem. 1991,63, 1719-1727. (8)Kaneyasu, M:; Ikegami, A.; Arima, H.; Iwanaga, S. IEEE Tram. Components, Hybrids Manufacturing Tech. 1987, CHMT-10,267-273. (9) Nakamoto,T.;Fukunishi,K.; Moriizumi, T. Sens. Actuators 1990, 1,473-476. (10) Nakamoto, T.; Fukuda, T.; Moriizumi, T. Sem. Actuators 1991, 3,1-6. 0003-2700/93/0365-0673$04.00/0
One approach to developing such a sensor is to use the human sense of smell as a model, as it can be used to suggest how the sensor arrays, transducers, and pattern recognition method should be designed. The human sense of smell has long been a topic of scholarly discussion. Attempts to define a theory of odor perception have been suggested in the fields of psychology, physiology, biology, and chemistry. The human sense of smell is the least understood of all human senses because odors are described only in subjective terms (there are no definitive words for odors). This might be attributable to the fact that the olfactory sense is more instinctive than other senses and is based on appetite and sex drive.“ The study of odor evaluation in terms of human reactions to them is important in relation to response data processing in an olfactory sensor. On the other hand, it appears biologically that odor reception is an organic interaction between olfactory receptor cella and odor molecules, followed by the formation of a nervous pattern (across-fiberpattern) which consists of the different responses of receptor cells to an odor. The pattern is transmitted to the brain and recognized.12 Many efforts have also been made to relate the recognized odor quality to the chemical parameters of odor molecules, for example, absorption spedra,ls molecular shape and size,14 electronic nature,16 gas chromatographic retention time indices,’&le and 1-octanoVwater partition coefficients.lQ These indicate that the interaction between organic materials and odors as well as pattern recognition techniques are important for developing an olfactory sensor. Based on the above factors, piezoelectric sensor arrays coated with organic compounds are attractive because the adsorbate detection and pattern recognition are similar to those in humans. In this work, 37 synthetic fragrances representing typical odors were evaluated subjectively by eight women who graded them on a seven-pointscale. Their reactions were quantified by using principal component analysis (human responses). A set of eight piezoelectric sensors incorporating different polymer materials was fabricated in order to investigate the adsorption of the same fragrances in the gas phase. The adsorption properties were examined by the application of cluster and principal component analysis to the sensor responses. Finally, the sensor responseswere correlated with the human responses by multiple regression analysis. The (11) Engen, T. The Perception of Odors; Academic Press: New York, 1982. (12) Takagi,S.F.Human Olfaction;UniversityofTokyoPreee: Tokyo, 1990. (13) Wright, R. H. J. Theor. Biol. 1977, 64, 473-502. (14) Amoore,J. E. Molecular Basis of0dor;C.C. Thomas: Springfield, 1970. (15) McGill, J. R.; Kowalaki, B. R. Anal. Chem. 1977,49, 696-602. (16) Laffort, P.;Patte, F.;Etcheto, M. Ann. N.Y. Acad. Sci. 1974,237, 193-208. (17) Lawrence, S.; Jura, P. C. Anal. Chem. 1990,62,2676-2684. (18) Edwards, P. A.; Anker, L. S.; Jurs, P. C. Chem. Sene. 1991, 16, 447-465. (19) Yoshii, F.;Liu, Q.;Hirono, S.;Moriguchi, I. Chem. Sem. 1991,16, 319-328. 0 1993 American Chemlcal Soclety
874
ANALYTICAL CHEMISTRY, VOL. 65, NO. 6, MARCH 15, 1993
Table I. Synthetic Fragrances Used in This Work notes fragrances (1)anisic aldehyde, (2) anethol anise (3) heptanal, (4) octanal, (5) nonanal aldehydic (6) undecanal, (7) dodecanal, (8) 2-methylundecanal (9) cinnamyl acetate, (10)methyl balsam cinnamate (11)isoamyl salicylate, (12) isobutyl salicylate citrus (13) citral, (14)citronellal, (15) citronellyl acetate (16)cinnamic aldehyde, (17) eugenol, spicy (18)methyleugenol (19)phenylpropyl alcohol leafy (20)nerol fresh floral (21) linalool, (22) terpineol (23) (phenylethy1)dimethylcarbinol (PEDC),(24) PEDC acetate (25) isoamyl acetate fruity jasmine (26) benzyl acetate, (27) benzyl formate, (28)benzyl propionate lavender (29) terpinyl acetate (30) linalyl acetate orange (31)geranyl acetate, (32) citronellol, rose (33)geraniol (34)phenylethyl alcohol, (petals) (35) phenylethyl acetate (36)geranyl formate, (37)geranyl (fruity) butvrate ~
Table 11. 55 Descriptive Characteristics Used in the Exmriment of Human Reactions to Fragrances 20. grassy 38. sultry 1. like medicine 21.sour 39. vital 2. fruity 22. minty 40. unhealthy 3. metallic 23. delicate 41. dislike 4. fatty 42. clingy 24. sporty 5. sweet 25. mild 43. irritating 6. spicy 26. youthful 44. clear 7. fragrant 27. elegant 45. dark 8. sexy 46. cheerful 28. restful 9. fresh 47. pleasant 29. muddy 10. feminine 48. soft 30. bright 11.powerful 49. white image 31. gloomy 12. heavy 50. brown image 32. tough 13. warm 51. purple image 33. dispirited 14. calm 52. green image 34. frank 15. refreshing 53. pink image 35. light 16. healthy 54. black image 36. cold 17. floral 55. gray image 37. rough 18. leafy 19. earthy
result showed the possibility of developing an artificial olfactory sensor.
EXPERIMENTAL SECTION The 37 typical synthetic fragrances (Yamamoto Perfumery Co. Ltd.) used in this work are listed in Table I. They are all commercially available and were used without further purification. Each fragrance (1mL) was put in a brown glass tube (5 mL) to prevent its color from influencing the human reactions. The test subjects were eight Japanese women in their twenties, all of whom live in Ibaraki prefecture, north of Tokyo. They smelled each fragrance and graded it on a seven-point scale in terms of 55 descriptive characteristics such as associations with color and emotion. These characteristics, translated into English, arelisted in Table 11. Onerandomly selectedfragrancewas graded by each woman each day. The results were analyzed by principal component analysis individually. The piezoelectric sensor consisted of a quartz crystal (10MHz AT-cut) with a Au (800 A)/Cr (50 A) electrode coated with a hydrophobic polymer film (0.8 Nm thick). The commercially purchased coating polymers (General Science Co. Ltd.) were polyethylene (PE),poly(2,6-dimethyl-p-phenyleneoxide) (PPO),
Organic Polymer Film Quartz Crystal I Electrode
m' ,
I W"
/
' /
Sensitivity;lHung
