PHYSICO-CHEMICAL AND ELECTRONIC NOSE MEASUREMENTS ON THE STUDY OF BISCUIT BAKING KINETICS

Abstract

The main physico-chemical modifications during the cooking process of laboratory-made biscuits were monitored at different cooking times (0, 2, 4, 6, 8, 10, 12 min). Moisture content, surface colours and textures were measured. In addition the evolution of the flavour release was performed by means of an electronic nose equipped with 10 metal-oxide sensors. Multivariate statistical analyses were performed to distinguish samples as a function of their physico-chemical characteristics. The electronic nose permitted differentiation between raw, under cooked, well-cooked and over cooked products. Similar and complementary information was obtained considering both electronic nose data and traditional physico-chemical cooking indexes. The obtained results showed that it could be feasible to monitor the changes in the biscuit's aroma and cooking level directly during the process by using an electronic nose with a simplified gas sensor array, as well as allowing the optimization of the technological parameters. [PUBLICATION ABSTRACT]

Headnote

ABSTRACT

The main physico-chemical modifications during the cooking process of laboratory-made biscuits were monitored at different cooking times (0, 2, 4, 6, 8, 10, 12 min). Moisture content, surface colours and textures were measured. In addition the evolution of the flavour release was performed by means of an electronic nose equipped with 10 metal-oxide sensors. Multivariate statistical analyses were performed to distinguish samples as a function of their physico-chemical characteristics. The electronic nose permitted differentiation between raw, under cooked, well-cooked and over cooked products. Similar and complementary information was obtained considering both electronic nose data and traditional physico-chemical cooking indexes. The obtained results showed that it could be feasible to monitor the changes in the biscuit's aroma and cooking level directly during the process by using an electronic nose with a simplified gas sensor array, as well as allowing the optimization of the technological parameters.

Key words: biscuit, colour, cooking process, electronic nose, texture

(ProQuest: ... denotes formulae omitted.)

INTRODUCTION

Biscuits are a popular food product, consumed by a wide range of people, due to their varied taste, long shelf-life and relatively low cost (VITALI et al, 2009).

The main sequential operations in the production of plain, round biscuits are ingrethent metering, dough mixing, dough sheeting, dough sheet relaxation, biscuit shape forming, baking, cooling and packaging. Each of the above mentioned steps in the process is of equal importance in determining the final character of the biscuit (CRONIN and PREIS, 2000).

Short doughs are characterised by their composition based upon three major ingrethents: flour, sugar and fat. Physico-chemical changes occurring in biscuit dough during the baking stage are very complex and take place following specific kinetics depending on heating conditions. The main transformations induced by cooking are water evaporation, protein denaturation, starch gelatlnisation/destruction, browning and surface colouration (principally due to the Maillard reactions), dough expansion during the production and the thermal expansion of gas cells formed during mixing (CHEVALLIER et al, 2000; MANLEY, 2001).

In the baking process, the viscoelastic dough is transformed into a solid-like baked item. This process determines the biscuit's final physical characteristics including dimensions (diameter and thickness), weight and moisture content (CRONIN et al., 2000). During the cooking process a typical evolution in the aroma profile takes place. In particular in the initial stages heating produces an increase in the volatility of the distinctive aromas of raw food, deriving from the ingrethents used in the formulation. Thereafter, cooking involves the genesis and release of new volatile compounds that are a consequence of chemical reactions occurring in the food matrix. Finally pirolysis reactions (i.e. Maillard and Strecker degradation) take place on the food's surface with the formation of specific volatile compounds (WARD et al, 2002). The biscuit's initial aromatic profile, due to the raw matrix composition, changes considerably becoming, at the end of cooking process, very complex in terms of the type and quantity of the numerous chemical substances.

Consumers consider texture and related flavour release as factors of primary importance in determining the profile of dry bakery products and assess texture by using a combinations of bites, particle decay and sound evaluations together with flavour release in the mouth (PIAZZA et al, 2008).

The complexity of most food aromas makes them difficult to characterize with conventional flavour analysis techniques such as gas chromatography or gas chromatography olfactometry (PERIS and ESCUDER-GOLABERT, 2009).

Food organoleptic features are usually assessed by human sensory methods. The classical techniques of descriptive analysis, namely flavour profile, quantitative descriptive analysis and texture profile, are mainly useful for the sensory characterisation of food products. These sensory methods need a group of well-trained assessors and several established attributes in order to provide reliable results; requirements that are time-consuming and, in some cases, could cause serious problems to some industries and laboratories (SINESIO et al, 2000). Hence, an instrument such as the electronic nose (e-nose), with recognised high sensitivity and the ability to provide data closely linked to results obtained from human sensory panels would be very useful for several specific applications in food control.

Because they are easy to build, cost-effective and as they provide analysis quickly, e-noses are becoming more and more popular as objective automated non-destructive techniques to characterize and /or monitor food flavours (PERIS et al, 2009).

GARDNER and BARTLETT (1994) defined the enose as "an instrument which comprises an array of electronic chemical sensors with partial specificity and an appropriate pattern recognition (PR) system capable of recognizing simple or complex odours".

E-noses were proposed some years ago as a promising technology for odour detection and discrimination, and it has been successfully applied in different fields, such as food science, medicine or environmental pollution control.

Relevant recent literature shows that there are five major categories of use for e-noses in food control. These are (i) process monitoring, (ii) shelf-life investigation, (iii) freshness evaluation, (iv) authenticity assessment and (v) other quality control studies (PERIS et al, 2009).

Detection of key aromas of bakery products during baking is of crucial importance at an industrial level with regard to the final food quality. Previous works showed the possibility of using an e-nose to detect and monitor the evolution of volatile compounds during the cooking process of some foods (WARD et al, 2002) or key aromas of bread baking (PONZONI et al, 2008). To our knowledge, there are no works in which the evolution of biscuit quality characteristics during baking has been monitored considering both traditional cooking indexes and a gas sensory array such as e-nose.

The aim of this work was to monitor the baking process of biscuits in terms of the main physico-chemical product characteristics (moisture content, colour and texture) together with the evolution of aromatic compounds during cooking time. The analysis of flavour release was performed by using of an e-nose equipped with 10 metal-oxide sensors. The obtained e-nose responses were subjected to a multivariate statistical analysis in order to classify the biscuits on the basis of their cooking level.

MATERIALS AND METHODS

Biscuit preparation

Biscuits were prepared using wheat flour, sugar, butter, eggs, baking powder and salt according to the formula given in Table 1 . The recipe was based on previous trials performed on the same materials.

The ingrethents were mixed in a household mixer (Kenwood, Major, Hampshire - UK) for 10 min, and, after mixing, the dough was kept for 20 min at room temperature (23° ± 1°C) in a closed food container in order to avoid dehydration. Afterwards, the dough was sheeted to a thickness of about 2.5 mm and then was cut by using a stainless steel circular mould (3.5 mm diameter) and placed on a tray. Biscuits were baked in an electric thermo-convection oven (FC61, ANGELO PO, Grandi Cucine S.p.A, Carpi, Italy) at 175°C for different cooking times: 2, 4, 6, 8, 10, and 12 min. The cooking times were chosen on the basis of preliminary experiments in order to obtain under and over baked products. The baking experiments were carried out in triplicate for each cooking time. For each test 16 biscuits were cooked in order to obtain enough samples for all the other measurements.

After cooking, the biscuits were removed from the oven and immediately set for the e-nose analysis. Conversely, before being used for the physico-chemical analyses (moisture content, texture and colour) the biscuits were cooled at room temperature for 1 hour.

Analyses of biscuits

After each baking test, the following analytical deterTninations were performed: moisture, colour, texture properties, and the flavour release.

Moisture

Biscuit moisture was determined by following the ICC Standard Method n° 110/1 (ICC 2004). For each sample three replicates were performed.

Colour analysis

by computer vision system (CVS)

Six biscuits from each cooking experiment, representative of whole sample, were used for colour measurements. Biscuits were placed on a matte black background and images were captured by using the image acquisition system developed by MENDOZA and AGUILERA (2004) with slight modifications. Samples were illuminated by using two parallel lamps (with two fluorescent tubes by lamp, model TLD Deluxe, Natural Daylight, 18W/965, Philips, NY, USA) with a colour temperature of 6500 K (D65, the standard light source commonly used in food research) and a colour -rendering index (Ra) close to 90%. A colour digital camera (CDC) mod. PowerShot A70 (Canon, NY, USA) was located vertically over the sample at a distance of 12.5 cm. Lamps and CDC were inside a wooden box with internal walls painted black in order to avoid the light and reflection from the room. Images of biscuit surfaces were taken on the matte black background and saved in JPG format. The algorithms for pre-processing of full images, image segmentation and colour quantification were written in MATLAB 6.5 (The MathWorks, Inc., USA). The average value of the segmented pixels in CIE L*a*b* colour space was registered as the colour of the sample. The biscuits' colour were described in terms of: luminosity (L*), hue angle (h°) = tan1(b*/a*) and Chroma ... (McGUIRE , 1992).

Texture

The mechanical and fracture properties of biscuit samples were determined by the three-point bending test (TYAGI et al., 2007) using a TA. HDi 500 Texture Analyser (Stable Micro Systems, Godalming, UK) equipped with a 5 kg load cell.

For the analysis, each biscuit was placed on the aluminium platform situated on two supporting beams spaced at a distance of 35 mm. The blade was attached to the crosshead of the instrument and was brought down to break the biscuit at a crosshead speed of 3.0 mm/s. Care was taken to see that the point of contact was at an equivalent distance from both the supporting beams. This test simulates the biscuit hardness during consumer handling and the way it breaks by bending (TYAGI et al, 2007). All the tests were carried out at room temperature. The blade was brought down at a constant speed under computer control and the applied force was recorded as a function of time. The absolute peak force from the resulting curve was considered as biscuit hardness, while the distance at the break (resistance of the sample to bending) was recorded as sample fracturability. Samples that break at a very short distance have a high fracturability. In order to better understand the resuits in this study the fracturability was expressed as 1 /break point distance. Six biscuits for each cooking time were tested.

E-nose and test procedures

Samples were analyzed as follows: about 3 g of powdered biscuit sample was placed into a glass vials (capacity 40 mL); these vials were then sealed with a suitable lid equipped with a pierceable Silicon /Teflon disk in the cap. Preliminary experiments, carried out in order to obtain signals of suitable intensity and good reproducibility, showed that after 60 min at 25° ± 1°C of equilibration, the headspace reached a steady state. After this period, the headspace was analyzed with a commercial portable e-nose PEN2 (Alrsense Analytics, Milano, Italy) composed of an array of 10 temperature-moderated metal-oxide sensors (MOS), a sampling system, a data acquisition device and a data processing system. The sensor response was expressed as relative conductivity (G /G0) of a single sensor with time, G0 is the initial conductivity taken as a reference.

The signal output of the sensors was digitized by recording, and normalized to a value of 1.0 prior to sampling; this arbitrary baseline value was subtracted from the sensor responses prior to enhancement determination. The signal output was measured at 1 s intervals for 60 s at a flow rate of 400 mL min1. The injection time was long enough for most of the sensors to reach a steady state condition. After each sample analysis the system was purged for 120 s with filtered air prior to the next sample injection to allow re-establishment of the instrument base line. No sensor drift was experienced during the measurement period. The sensor values from 40 to 60 s were used for the statistical analysis.

For each sample, 3 different vials were prepared and each vial was evaluated three times; an average of these nine measurements for each cooking time was used for the statistical analysis.

Statistical analysis

The pattern recognition techniques used in this work were Principal Component Analysis (PCA) and Cluster Analysis (CA). PCA was used to reduce the number of variables in the data matrix and to select the most discriminating parameters in order to classify the different samples. CA was applied on electronic-nose selected sensor responses. CA performs agglomerative hierarchical clustering of objects on the basis of distance measurements of dissimilarity or similarity.

The statistical package STSG statistica for Windows, version 6.0 (Stat-Soft Inc., Tulsa, OK, USA) was used.

RESULTS AND DISCUSSION

Baking is the key stage in baked good production; during this process the main quality characteristics of the final product such as size extension, brownness, texture and flavour are formed thanks to several physico-chemical changes in the product. During baking, transformations, which depend on the course of water content and temperature, are decisive for the final product quality. The majority of these transformations occur together and influence each other (HADIYANTO et al., 2007).

In the first part of this study, the cooking kinetics of biscuits were analyzed on the basis of the main changes regarding the most important parameters involved in the phenomenon.

In Fig. 1, the changes of the biscuit's water content as a function of baking time are shown.

It can be observed that at the beginning of cooking there is a faster reduction in the level of moisture owing to the water migration from the inner towards the biscuit surface due to diffusion and capillarity phenomena.

After 8 min, the optimum cooking time on the basis of previous experiments, the moisture loss rate decreased reaching the minimum and constant value after 10 min.

Usually, in the last stage of baking, when most of water has evaporated, a dried and coloured crust starts to form.

Colour is one of the most important quality attributes for consumer acceptance of baked goods. In fact the surface colour of a baked product is one of the main criterion of acceptance by consumers, together with texture and taste.

Colour is influenced by dough composition, water content and process conditions, temperature and duration of baking. Sugars together with proteins will produce browning compounds, which give colour to the product that lead to irreversible changes. The required relative high temperature needed for these reactions is reached only in the outer zones, once the product is sufficiently dried to have a low water activity and corresponding high evaporation temperature. The duration of this phase depends on the initial water content of the dough and is ruled by heat and mass transfer phenomena (HADIYANTO et al., 2007).

In Fig. 2 colour results (lightness (L*), hue angle (h0) and Chroma (C*)) measured on the biscuit samples at different cooking times are reported. Obtained variations of colour parameters during baking draw a characteristic trajectory, previously found in similar researches (BROYART et al., 1998; CHEVALUER et al, 2002).

During the first 4 min of cooking the lightness (L*) and hue angle (h°) values increased respectively from about 70 to 80 and from 87 to 90, while chroma (C*) decreased from about 46 to 39. Subsequently, with the increase of the product's temperature, lightness started to decrease and the biscuit surface started to brown. These findings suggest that the biscuit samples underwent initial whitening and a subsequent gradual intensification of colour, from a yellow to a brown hue. After 8 min of baking the considerable decrease in lightness, hue and chroma were concomitant with a dark browning of the biscuit surface (Fig. 3), that becomes unacceptable for consumers.

As is known and established by previous researches (SHIBUKAWA et al, 1989; BROYART et al, 1998; CHEVALLIER et al., 2002) the overall biscuit colouring kinetics involve two stages. During the first stage of baking the increase in lightness, hue and decrease in chroma correspond to a brightening of the biscuit surface; this makes the product surface appear to be lighter. These colour changes are due to the water migration toward the surface and /or modification in the surface state attributable to the sample rise (CHEVALLIER et al, 2002). In the second stage the progressive browning of the biscuit surface is mainly caused by chemical reactions (Maillard reactions, sugar caramelisation) activated by thermal treatment.

In Fig. 4 the changes of texture parameters of biscuit samples at different cooking times are shown.

As expected, the longer the cooking time the higher the hardness and the fracturability of the biscuits. During cooking, biscuit samples underwent an increase in hardness values from 0.46 kg after 2 min of cooking to 2.21 kg at the end of the process. Regarding the fracturability, the biscuits exhibited values between 1.29 (1/mm) after two min of cooking and 3.76 (1/mm) at the end of the process. The rate of fracturability increase was faster until the first 8 min of cooking; after this time, when the evaporation of water and the production of gases had ended, the fracturability increased more slowly.

These macroscopic changes of biscuit texture characteristics are caused by several phenomena that take place during baking: the production of gases from chemical liveners, the water vaporization, the gas thermal expansion associated with a biscuit's thickness increases and the product dries, resulting in a large decrease in product density and on the development of an open porous structure (CHEVALLIER et al., 2002).

As previously reported, during cooking changes in the biscuit's colour and texture are concomitant to the evolution of a typical aroma profile.

In Fig. 5 the responses (relative conductivity mean data of six replicates) of the e-nose sensors (MOS 1 to 10) obtained at each considered biscuit cooking time are shown.

It can be observed that only some sensors were more sensitive to the evolution of biscuit aromatic profile during baking. In particular, the responses of 1 , 3 and 5 sensors increased with the increasing of cooking time; the 2 and 6 sensors displayed the highest relative conductivity signals. From these results, it seems that the responses of the 1 , 3 and 5 sensors are most related to the aroma compounds that are formed and developed during the cooking process. Meanwhile, the 2 and 6 sensor responses seem to be related to the disappearance and/ or conversion of others chemical volatile compounds. All the other sensor results were less sensitive to the aromatic profile developed during biscuit cooking process.

PCA was carried out by using the responses from the e-nose sensors collected at the end of each cooking time, in order to highlight and to evaluate whether the most sensible e-nose sensors were able to detect differences between the biscuit samples during cooking stages as a function of their volatile components. The PCA loading plot reported in Fig. 6 confirms results shown in Fig. 5.

Few sensors were able to explain most of the variance and in particular the MOS 1, 3, 5, 2 and 6 had the highest influence in the pattern file. Their responses were the most relevant in the discrimination of biscuits on the bases of the level of cooking. The 4 and 7 MOS sensor responses were not significant from the PCA. Examining the score plot of Fig. 7 in the area defined by the first two principal components (96.21% of the total variance) a clear separation of biscuit samples into different clusters, according to the cooking degree was found.

Moving from right to left along the PC 1 axis (79.08% of the total variance), raw biscuit cluster is first encountered, then the biscuits were cooked for 2, 4, 6, 8, 10 and 12 min respectively. In particular, the group of sensors on the right of the loading plot better discriminate the raw and under cooked (0-4 min) biscuit aromatic compounds; while those on the left of the plot (MOS 1, 3 and 5) better discriminate the cooked and over cooked samples. The biscuit samples cooked for 6 and 8 min were grouped in the same cluster, probably because of their similar aromatic profiles. The biscuits cooked for 10 min were positioned among them and the over cooked (10 to 12 min) samples. The obtained finger print (sequence) shows a behaviour similar to that obtained by PONZONI et al (2008) who detected, with only four metal oxide e-nose sensors, key artificial aromas related to different stages of the bread baking process. In the same way at which biscuit samples were here distinguished and separated along the PC axes, they firstly obtained the separation of acetaldehyde and diacetyl, that are key odorants identifying the initial baking steps, and then of pyridines and pyrazines, that are flavouring compounds produced by Maillard reactions as the baking process continues. To verify results obtained from PCA, a cluster analysis was run with the same e-nose sensor selected responses (Fig. 8) in order to isolate groups of biscuits with similar aromatic profiles. The dendogram was obtained using Ward method and Euclidean distance.

Through this analysis, it was possible to subdivide the biscuit samples on the bases of their aromatic profiles, related to cooking times, into five groups: 0 (raw), 2, 4, 6 and 8 min, 10 and 12 min. The distances among biscuit groups increased proportionally to the increase of cooking time, being characterized by more and more different volatile aroma compounds.

The CA results were in agreement with those obtained from PCA and were able to better distinguish the biscuit samples cooked for 10 and 12 min from the others. These outcomes corifirrn further the suitability of the e-nose in monitoring the evolution of the biscuit aroma changes and cooking degree.

CONCLUSIONS

The potential of the e-nose, as an off-line system, for distmguishing the biscuits' cooking levels on the bases of their peculiar aromatic profiles was demonstrated. Multivariate statistical analysis of the e-nose data collected during the biscuit cooking process showed a clear separation between raw, under-cooked, well-cooked and over-cooked biscuits. The discrimination of biscuits according to the cooking time, obtained by analyzing the e-nose signals, confirmed the results provided from all the other considered traditional cooking indexes (moisture content, colour and texture), showing the interchangeable ability of both approaches.

In particular, only few sensor responses were able to explain most of the variance of the data in the PCA and consequently were relevant in the discrimination of the biscuits' cooking levels.

The use of e-nose would appear to have a great potential in the future to develop an online monitoring system of the baking process. For this purpose much more research is required, as well as many calibration tests to demonstrate its on-line application.

References

REFERENCES

Broyart B., Tiystram G. and Duquenoy A. L 998. Predicting colour kinetics during cracker baking. J. Food Eng. 35: 351.

Chevallier S., Colonna P., Della Valle G. and Lourdin D. 2000. Contribution of major ingrethents during baking of biscuit dough systems. J. Cereal Sci. 31: 241.

Chevallier S., Della Valle G.P., Colonna B., Broyart B. and Tiystram G. 2002. Structural and chemical modifications of short dough during baking. J. Cereal Sci. 35: 1 .

Cronin K. and Preis C. 2000. A statistical analysis of biscuit physical properties as affected by baking. J. Food Eng. 46: 217.

Gardner J.W. and Bartlett PN. 1994. Abrief history of electronic noses. Sensor Actuat B- Chem. 18: 211.

Hadiyanto H., Asselman A., Van Straten G., Boom R.M., Esveld D.C. and Van Boxtel A.J.B. 2007. Quality prediction of bakery products in the initial phase of process design. Innov. Food Sci. Emerg. Technol. 8: 285.

ICC. 2004. "International Association for Cereal Science and Technology". Determination of the moisture content of cereals and cereal products (Practical method). Standard Method 110. Approved 1960, revised 1976. Vienna.

Manley D. 200L. Baking Techniques. In: Manley D. (Ed.), "Biscuit, cracker and cookie recipes for the food industry", p. 27. Woodhead Publishing Ltd and CRC Press LLC, Cambridge England.

Mcguire R.G. 1992. Reporting of objective colour measurements. Hort. Sci. 27: 1254.

Mendoza F. and Aguilera J. M. 2004. Application of Image Analysis for classification of ripening Bananas. JFS E: Food Eng. Physical Prop. 69: E471.

Peris M. and Escuder-Golabert L. 2009. A 2 1st century technique for food control: Electronic noses. Anal. Chim. Acta 638: 1.

Piazza L., Gigli J. and Benedetti S. 2008. Study of structure and flavour release relationship in low moisture bakery products by means of the acoustic-mechanical combined technique and the electronic nose. J. Cereal Sci. 48: 413.

Ponzoni A., Depari ?., Faiasconi M., Comini E., nammini ?., Marioli D., Taroni A. and Sberveglieri G. 2008. Bread baking aromas detection by low-cost electronic nose. Sensor Actuat B-Chem. 130: 100.

Shibukawa S., Sugiyama K. and Yano T 1989. Effect of heat transfer by radiation and convection on browning of cookies at baking. J. Food Sci. 54: 621.

Sinesio F, DI Natale C, Quaglia G.B., Bucarelli FM., Moneta E., Macagnano A., Paolesse R. and DAmico A. 2000. Use of electronic nose and trained sensory panel in the evaluation of tomato quality. J. Sci. Food Agrie. 80: 63.

Tyagi S.K., Manikantan M.R., Oberoi H.S. and Kaur G. 2007. Effect of mustard flour incorporation on nutritional, textural and organoleptic characteristics of biscuits. J. Food Eng. 80: 1043.

Vitali D., Vedrina Dragojevic I. and Sebecic B. 2009. Effects of incorporation of integrai raw materials and dietary fibre on the selected nutritional and functional properties of biscuits. Food Chem., 1L4: L462.

Ward D., Benedetti S. and Riva M. 2002. Monitoring and controlling coking processes using an electronic olfaction device, pp. 292-8. Proceedings of the 9th International Symposium on Olfaction and Electronic Nose (ISOEN); 2002 Rome, Italy.

Paper received March 8, 2011 Accepted September 8, 2011