Meat Science 65 (2003) 1175–1185 www.elsevier.com/locate/meatsci
Detection of Iberian ham aroma by a semiconductor multisensorial system Laura Otero, Ma Carmen Horrillo*, Marı´a Garcı´a, Isabel Sayago, Manuel Aleixandre, Ma Jesu´s Ferna´ndez, Luis Are´s, Javier Gutie´rrez Laboratorio de Sensores, Instituto de Fı´sica Aplicada (IFA)–Consejo Superior de Investigaciones Cientı´ficas (CSIC), Serrano, 144, E-28006, Madrid, Spain Received 10 July 2002; received in revised form 9 December 2002; accepted 9 December 2002
Abstract A semiconductor multisensorial system, based on tin oxide, to control the quality of dry-cured Iberian hams is described. Two types of ham (submitted to different drying temperatures) were selected. Good responses were obtained from the 12 elements forming the multisensor for different operating temperatures. Discrimination between the two types of ham was successfully realised through principal component analysis (PCA). # 2003 Elsevier Ltd. All rights reserved. Keywords: Hams; Gas sensors; Electronic nose; Volatile compounds; Aroma detection
1. Introduction Dry-cured Iberian ham is a product of great economic importance in Spain and is expanding its markets in Northern Europe, United States and Canada (Gonza´lez Blasco, 2001). Its intense and characteristic flavour make it a delicacy and justifies its higher price compared with other kinds of cured hams. Ham curing is an extensive process, over 24 months in some cases, with two definitive steps: salting-post salting and ripening. During the process, temperature and humidity are accurately controlled to reduce the risk of bacterial spoilage. The quality of the ham is determined towards the end of the curing process by a skilled worker. This person inserts a single probe into the ham and sniffs it with the odour exuding from the ham. However, the process is tedious and the quality of the hams cannot be monitored at frequent intervals during the curing process. Of the total final production volume, 1–2% results in spoiled hams that have developed undesirable flavours due to putrefaction reactions that take place during ripening. This is a great economic loss because of the complexity and length of the curing pro* Corresponding author. E-mail address:
[email protected] (Ma Carmen Horrillo). 0309-1740/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0309-1740(02)00347-9
cess that could be avoided by an earlier quality determination. The aroma of Iberian dry-cured ham is its most important quality parameter and is markedly affected by the raw material and the technology (time, temperature and relative humidity during the different phases of the curing process). The aroma is the result of both enzymatic and non-enzymatic reactions and degradations of macromolecules in the tissues of the hams during the drying and ripening periods. In addition to these pathways, secondary metabolism of microorganisms are involved in the development of volatile aroma components (Dirinck & Van Opstaele, 1998). So, the knowledge of the odorous compounds at each stage of processing is very important for control of the quality of the food product. Until now, most studies on the aroma components of hams have been carried out with sophisticated and expensive instrumentation such as gas-phase chromatography–mass spectrometry combined with extraction techniques such as high vacuum distillation and dynamic headspace analysis. Such procedures identified 77 compounds in the volatile fraction, mainly alkanes, branched alkanes, aldehydes and aliphatic alcohols. Small amounts of lactones, esters and ketones and other miscellaneous compounds are also present (Barbieri et al., 1992; Garcı´a, Berdague´, Antequera, Lo´pez-Bote, Co´rdoba, & Ventanas, 1991; Sabio,
1176
L. Otero et al. / Meat Science 65 (2003) 1175–1185
Vidal-Arago´n, Bernalte, & Gata, 1998). The great shortcoming of these techniques is that they are not able to measure the components in real time and continuously. Furthermore, analysis time is very long, so that they cannot be used routinely. In recent years, sensor arrays (electronic noses) have been increasingly used for the determination of food and beverage quality (Schaller, Bosset, & Escher, 1998). These kinds of systems have been successfully assayed in a pilot plant study of Serrano hams to assess, noninvasively, the quality at an early stage of the curing process (Abass et al., 1999). The objective of this paper was to test the sensitivity to ham aroma of different thin tin oxide gas sensors in order to integrate them, in the future, into an electronic nose able to discriminate different kinds of hams (spoiled or not, Iberian or not, . . .).
2. Material and methods 2.1. Multisensor fabrication The multisensor includes 12 sensor elements formed by thin tin oxide layers distributed radially on an alumina substrate of 1 inch diameter. Tin oxide was grown by sputtering (radio frequency type and magnetron mode) at 250 C using a SnO2 target under a 10% oxygen–90% argon mixture at a total pressure of 0.5 Pa. Some sensors were doped with different amounts of Pt, Pd and TiO2 by changing the deposition time during the process of sputtering. Pt and Pd dopants were introduced from metallic targets as an intermediate discontinuous layer between two tin layers with the same thickness. TiO2 was deposited on a tin oxide layer as a continous layer over it. Electrical contacts of Pt were also deposited by sputtering. The characteristics of each sensor are listed in Table 1, the doping levels being expressed as sputtering times (seconds).
The multisensor device was placed in a stainless steel test chamber and resistance measurements were carried out under a constant synthetic air flow of 200 ml/min (obtained from commercial pure gases: 78% N2 and 21% O2). The sensors were stabilised in dry air before their exposure to ham vapours during 30 min. 2.2. Samples Samples were extracted from Biceps femoris muscles of two different kinds of dry-cured hams (ready-to-eat): hams dried at a maximum temperature of 20 C and hams dried at a maximum temperature of 30 C (period in dryer: 3–4 months). Temperature in the drying period of ham curing is a key factor in the quality of the final product due to its influence on the enzymatic and nonenzymatic reactions that occur (Ventanas, Andre´s, & Garcı´a, 2000). Hams dried at 30 C had an intense and pleasant flavour caused by the abundance of amino acids, specially those that enhance a strong and umami flavour (delicious). On the other hand, hams dried at 20 C were sweeter, mainly due to the flavour of such free amino acids as alanine. All hams were from Iberian pigs. They were vacuum sealed and stored at 2 C until experimentation. 2.3. Extraction and detection of aroma from hams Five grams of minced ham were placed in a Drechsel bottle and kept at ambient temperature for 15 min to concentrate the aroma. During the headspace-generation step, a synthetic air flow of 200 ml/min was blown into the measurement chamber to create the baseline for each sensor. After 15 min, the aroma from the minced ham was forced by the synthetic air flow of 200 ml/min into the measurement chamber for 30 min. The electrical resistance of each sensor was measured in air (Ra) and in the sample (Rs) to evaluate the sample sensitivity (S) of each sensor defined as:
Table 1 Characteristics of the 12 sensors tested Sensor
Semiconductor material
1 2 3 4 5 6 7 8 9 10 11 12
SnO2 SnO2 SnO2 SnO2 SnO2 SnO2 SnO2 SnO2 SnO2 SnO2 SnO2 SnO2
(1 layer: 200 nm) (1 layer: 300 nm) (1 layer: 350 nm) (2 layers: 150 nm) (2 layers: 150 nm) (2 layers: 150 nm) (2 layers: 150 nm) (2 layers: 150 nm) (2 layers: 150 nm) (1 layer: 150 nm)+TiO2 (1 layer: 45 nm) (1 layer: 200 nm)+TiO2 (1 layer: 32.5 nm) (1 layer: 200 nm)+TiO2 (1 layer: 45 nm)
Dopant time (s) – – – Pt (4) Pt (8) Pt (16) Pd (4) Pd (8) Pd (16) – – –
L. Otero et al. / Meat Science 65 (2003) 1175–1185
Sð%Þ ¼
Ra Rs 100 Ra
ð1Þ
All the measurement steps were automatically controlled through standard IEEE interface boards by a personal computer. In each sensor, data were acquired at a rate of one measurement per 30 s. The response of the sensors to the different kinds of dry-cured hams was tested in the temperature range of 150–250 C and each experiment was duplicated. All the experiments at the same temperature were done on the same day and consecutive tests corresponded always to a different kind of ham. This experimental design allowed us to avoid possible effects of drift, a problem widely reported in the literature (Vernat-Rossi, Vernat, & Berdague´, 1996).
3. Results and discussion 3.1. Responses of the sensors in air The resistance change of the sensors for ham aroma depends on the resistance of the sensor at a working temperature in air before exposure to the sample gases. This resistance was different depending on the working temperature and the type of sensor (Fig. 1). In general, sensor resistance in air was minimal at 200 C reaching
1177
its maximum value at 150 C. Sensors fabricated with pure SnO2 (sensors 1–3) showed the lowest resistance in air over the whole temperature range. For all the temperatures assayed the thinest sensor (sensor no. 1) presented the biggest resistance in air. Adding Pt or Pd as dopants (sensors 4–6 and sensors 7–9, respectively) involved an increased resistance in relation to the same thickness of SnO2 (comparisons with sensor no. 2), the smaller deposition times giving a bigger resistance increment. Sensors with TiO2 onto the SnO2 layer also produced a resistance increase. 3.2. Responses of the sensors to Iberian ham aroma The 12 sensor elements in the multisensor were tested for their sensitivity to aroma from two different kinds of Iberian ham (dried at a maximum temperature of 20 C and dried at a maximum temperature of 30 C) in the range of 150–250 C. Results described in this section of the paper correspond to experiments using samples dried at 30 C. Responses from samples dried at 20 C showed identical trends. 3.2.1. Transient responses of the sensors Fig. 2 shows the typical transient responses (represented as sensitivity values) of one of the sensors (sensor no. 1) to the ham aroma at different temperatures. This kind of response results from complex phenomena of
Fig. 1. Resistance Ra (mean valuesstandard deviation), in ohms, of the twelve sensors tested in air at different temperatures.
1178
L. Otero et al. / Meat Science 65 (2003) 1175–1185
Fig. 2. Transient response of sensor no. 1 (pure SnO2 sensor of 200 nm) to ham aroma at different temperatures.
mass transfer and volatile compound–sensor interactions. These interactions involve the complex mechanisms of adsorption, desorption, competition and even of catalysis of the volatile compounds at the surface of the sensors (Morrison, 1982). Signals could be divided into three parts. The first, made up of the ascending zone of the curve, was due to the fact that volatile compounds concentrated in the headspace came to the surface of the sensors. The second part of the curve was a ‘plateau’ corresponding to an equilibrium between the steady-state desorption of the volatile compounds of the sample and its reaction with the sensible surface of the sensors (Vernat-Rossi, Garcı´a, Talon, Denoyer, & Berdage´, 1996). The descending part of the curve represents the reversibility of the sensor when the air flow was established. Temperature is the main factor that determines the shape of the transient responses. That is, the rise time and the recovery time became shorter with increasing temperature. Also the maximum values of sensitivity were reached at the higher temperatures. This is true for all the studied sensors. At 200 and 250 C, on exposure to ham aroma, the sensitivity values increased quickly, achieving 100% of the full change in 1 min, and recovered completely to the base line within 1 min when the air flow was resumed. When working at the lower temperature (150 C) some differences can be seen in the transient response of
the different sensors (Fig. 3). Pt and Pd doped sensors (sensors 4 and 7, respectively) gave slower responses and longer recovery times than the corresponding pure SnO2 sensor (sensor no. 2 with the same thickness). Also the maximum value of sensitivity was lower. The kinetic response of pure SnO2 and TiO2 doped sensors were, on the contrary, similar. 3.2.2. Maximum sensitivity values Fig. 4 shows the maximum sensitivity values (mean and standard deviation) to ham aroma for pure SnO2 sensors (sensors 1–3). Sensitivity to ham aroma increased with temperature, 250 C being the temperature of maximum response. The thinnest sensor (sensor no. 1) gave the maximum response at 150 and 200 C. At 250 C, differences in sensor responses were small. Figs. 5 and 6 show the maximum responses (mean and standard deviation) for Pt doped sensors and Pd doped sensors, respectively. In both figures, maximum responses of sensor no. 2 (pure SnO2 sensor with similar thickness) are also shown for comparison. Again, maximum sensitivity values were reached for the higher temperatures. All the Pt doping times gave similar responses. Doped sensors gave higher responses than sensor no. 2 at 200 and 250 C. In contrast, Pd doping times were influenced the sensor responses, the shorter doping times giving the maximum responses. Nevertheless, compared to pure SnO2 sensor no. 2, doping
L. Otero et al. / Meat Science 65 (2003) 1175–1185
Fig. 3. Transient responses of different sensors to ham aroma at 150 C.
Fig. 4. Maximum sensitivity valuesstandard deviation to ham aroma for SnO2 sensors (sensors 1–3) at different temperatures.
1179
1180
L. Otero et al. / Meat Science 65 (2003) 1175–1185
Fig. 5. Maximum sensitivity valuesstandard deviation to ham aroma for SnO2 sensors doped with Pt (sensors 4–6) at different temperatures. Sensor no. 2 of pure SnO2 with identical thickness is shown for comparison.
Fig. 6. Maximum sensitivity valuesstandard deviation to ham aroma for SnO2 sensors doped with Pd (sensors 7–9) at different temperatures. Sensor no. 2 of pure SnO2 with identical thickness is shown for comparison.
1181
L. Otero et al. / Meat Science 65 (2003) 1175–1185
with this element only increased its sensitivity at 250 C and only for the shorter doping time (sensor no. 7). Fig. 7 shows the maximum sensitivities (mean and standard deviation) of sensors doped with TiO2 (sensors 10–12). Maximum sensitivities of sensor no. 1 (pure
SnO2 sensor with the same thickness) are also shown for comparison. Sensor 10 had unusual behaviour, probably due to a manufacturing error. Its maximum response was reached at 200 C. Sensors 11 and 12 increased their sensitivity with temperature, sensor 12,
Fig. 7. Maximum sensitivity valuesstandard deviation to ham aroma for SnO2 sensors doped with TiO2 (sensors 10–12) at different temperatures. Sensor no. 1 of pure SnO2 with identical thickness is shown for comparison. Table 2 Meanstandard deviation values of maximum sensitivities obtained in the experiments Sensor
1
2
3
4
5
6
7
8
9
10
11
12
150 C Hams dried at 20 C Hams dried at 30 C 200 C Hams dried at 20 C Hams dried at 30 C 250 C Hams dried at 20 C
54.6 1.4 50.3 2.9 45.3 3.8 45.14.4 46.23.6 44.84.0
49.9 6.5
495.8 51.44.8 19.43.2
46 4.8 40.1 4.9 39.34.0 40.53.8 37.62.9 41.24.0
173.4 51.43.2 58.94.6
445.4 15.41.3 13.14.2
456.7 51.76.2
59.6 1.3 57.1 1.6 53.7 1.7 56.33.4 56.73.7 55.63.4 52.73.1 50.61.9 38.24.0 51.44.5 59.81.1 60.11.6
54.0 1.4 52.3 4.0 47.7 3.5 54.04.3 55.54.5 55.14.1 52.04.1 46.53.0 36.13.8 51.91.4 53.61.2 56.24.1
75.7 2.0 73.7 2.2 75.8 1.2 72.72.1 74.71.9 76.41.9 81.62.3 74.11.5 59.41.7 29.63.1 65.01.9 71.82
Hams dried at 75.6 1.8 73.9 1.9 75.8 1.4 76.32.5 74.91.7 75.41.1 79.71.5 74.01.7 59.51.2 33.27.6 67.11.3 71.12.2 30 C
1182
L. Otero et al. / Meat Science 65 (2003) 1175–1185
Fig. 8. Maximum responses at 150 C, expressed as sensitivity values, of the 12 sensors tested for aroma from two different kinds of Iberian hams (two different drying temperatures).
Fig. 9. Maximum responses at 250 C, expressed as sensitivity values, of the 12 sensors tested for aroma from two different kinds of Iberian hams (two different drying temperatures).
L. Otero et al. / Meat Science 65 (2003) 1175–1185
with the longer doping time, being the most sensitive. Its responses were better than those of sensor 1 with identical SnO2 thickness (200 nm) for low temperatures (150 C and 200 C). 3.3. Discriminating between different kinds of dry-cured ham Two types of dry-cured Iberian ham (according to the highest temperature, 20 C or 30 C, during the period in the dryer, 3–4 months) were tested. Table 2 shows the mean standard deviation values of the maximum sensitivities reached for both types of ham at the temperatures assayed. Fig. 8 compares these sensitivity values at 150 C. Responses for hams dried at 20 C were always
1183
higher than those corresponding to hams dried at 30 C. Nevertheless this behaviour was not the same for all working temperatures. With increasing temperature, differences in responses between both types of ham became smaller and at 250 C (Fig. 9) these differences were almost indiscernible. More information can be deduced from the transient responses of the sensors. Fig. 10 shows the transient responses of sensor no. 4 at 150, 200 and 250 C. At 150 C (Fig. 10a) curves corresponding to the two types of ham are perfectly separated and the maximum response is reached at the end of the sampling time. Responses for hams dried at 20 C were always higher over the entire period of sampling and this is true for all sensors. At 200 C (Fig. 10b), the maximum sensitivity
Fig. 10. Transient responses of sensor no. 4 (SnO2 sensor doped with Pt) at different temperatures for aroma from two different kinds of Iberian ham (two different drying temperatures). (a) 150 C. (b) 200 C. (c) 250 C.
1184
L. Otero et al. / Meat Science 65 (2003) 1175–1185
value was reached at the beginning of the sampling period and during it, curves represented in Fig. 10, almost overlap. In contrast, at 250 C (Fig. 10c), maximum responses corresponding to the two types of ham were close but the transient responses were markedly different those corresponding to hams dried at 30 C always being higher. This type of behaviour was the same for all sensors. The method of principal components analysis (PCA) has been used to discriminate the types of ham tested with regard to the operating temperature for 12 measurements (Fig. 11). Responses at the end of the sampling period have been employed. The principal factors 2 and 3 have been chosen because the principal factor 1 does not contribute to a clear discrimination between the types of ham. This can be due to an ageing process in the sensors which produces drift. In general, it can be observed that at any operating temperature both types
of ham are differentiated, better discrimination is obtained at 250 C. Fig. 12 shows the contribution of the sensors to the discrimination analysis. It is observed that sensors 1, 5, 6, 8, 9, 10 and 12 are the best to carry out the discrimination. Sensors 4, 5, and 11, 12 gave similar responses.
4. Conclusions All the sensors tested gave, in general, good responses to ham aroma, 250 C being the operating temperature at which the best sensitivities and discriminations are obtained. In a few cases dopant incorporation has improved the sensitivity to ham aroma. The use of semiconductor sensor systems can provide a non-destructive, non-contact method of food analysis, and would be useful for quality control in the food industry.
Acknowledgements This work has been carried out under the financial support of the European Commission (FEDER founds) and Spanish Science Commission (CICYT) (Project 1FD97-0723). The samples of ham were kindly supplied by the Department of Food Technology of the Faculty of Veterinary (University of Extremadura).
References
Fig. 11. PCA applied to two Iberian ham types at three operating temperatures: 150, 200 and 250 C.
Fig. 12. Loading plot from PCA.
Abass, A. K., Cooper, L. D., De Lacy Costello, B. P. J., Evans, P., Ewen, R. J., Hart, J. P., Ratcliffe, N. M., & Wat, R. K. M. (1999). A pilot plant study of an automated quality control electronic nose for the monitoring of dry cured hams. In Electronic noses and sensor array based systems, design and applications. Proceedings of the 5th International Symposium on Olfaction and the Electronic Nose, (pp. 211–224). Barbieri, G., Bolzoni, L., Parolari, G., Virgili, R., Buttini, R., Careri, M., & Mangia, A. (1992). Flavor compounds of dry-cured ham. Journal of Agriculture and Food Chemistry, 40, 2389–2394. Dirinck, P., & Van Opstaele, F. (1998). Volatile composition of southern European dry-cured hams. In E. T. Contis, et al. (Eds.), Developments in food science (food flavours: formation, analysis and packaging influences) (Vol. 40) (pp. 233–243). Amsterdam: Elsevier Science B.V. Garcı´a, C., Berdague´, J. J., Antequera, T., Lo´pez-Bote, C., Co´rdoba, J. J., & Ventanas, J. (1991). Volatile components of dry cured Iberian ham. Food Chemistry, 41, 23–32. Gonza´lez Blasco, J. (2001). Las denominaciones de origen (DO) e indicaciones geogra´ficas protegidas (IGP) de jamo´n reconocidas en la Unio´n Europea. Ca´rnica 2000, April, 83–94. Morrison, S. R. (1982). Semiconductor gas sensors. Sensors and Actuators, 2, 329–341. Sabio, E., Vidal-Arago´n, M. C., Bernalte, M. J., & Gata, J. L. (1998). Volatile compounds present in six types of dry-cured ham from south European countries. Food Chemistry, 61(4), 493–503. Schaller, E., Bosset, J. O., & Escher, F. (1998). ‘Electronic noses’ and their application to food. Lebnsm.- Wiss. u.- Technol, 31, 305–316.
L. Otero et al. / Meat Science 65 (2003) 1175–1185 Ventanas, J., Andre´s, A. I., & Garcı´a, C. (2000). Condiciones del procesado que favorecen el desarrollo del ‘‘flavor’’: Influencia de la sal, la temperatura y la duracio´n del proceso madurativo. In J. Ventanas et al. (Eds.). Tecnologı´a del Jamo´n Ibe´rico (pp. 321– 341), Mundi-Prensa. Vernat-Rossi, V., Garcı´a, C., Talon, R., Denoyer, C., & Berdague´, J. L. (1996). Rapid discrimination of meat products and bacterial
1185
strains using semiconductor gas sensors. Sensors and Actuators B, 37, 43–48. Vernat-Rossi, V., Vernat, G., & Berdague´, J. L. (1996). Discrimination de produits agroalimentaires par capteurs de gaz a` semi-conducteurs fonctionnant avec l’air de l’atmosphe`re ambiante du laboratoire. Diffe´rentes approches de traitement du signal. Analusis, 24, 309– 315.
