Sensors and Actuators B 84 (2002) 113±122
A semiconducting metal-oxide array for monitoring ®sh freshness Jeremy Hammonda,*, Brent Marquisa, Ray Michaelsa, Brian Oicklea, Bruce Segeeb, John Vetelinob,c, Al Bushwayd, Mary Ellen Camired, Kathy Davis-Denticid a
Sensor Research and Development Corporation, 17 Godfrey Drive, Orono, ME 04473, USA Electrical and Computer Engineering Department, University of Maine, Orono, ME 04469, USA c Laboratory for Surface Science and Technology, University of Maine, Orono, ME 04469, USA d Department of Food Science and Human Nutrition, University of Maine, Orono, ME 04469, USA b
Accepted 17 December 2001
Abstract An array of semiconducting metal-oxide (SMO) chemiresistive sensors can quantitatively measure the freshness of Atlantic salmon (Salmo salar), haddock (Melanogrammus aegle®nus), and Atlantic cod (Gadus morhua). A variety of SMO ®lms were tested, including ®lms containing oxides of copper and tin, and commercially available tin-based SMO ®lms. Analytical testing of the ®sh samples was performed in conjunction with SMO sensor testing of the volatile gases emitted from the degrading ®sh. Testing included the use of a sensory evaluation panel and tests incorporating amine colorimetric methods, pH analyses, and bacterial aerobic and anaerobic plate counts. Sensory analysis, trimethylamine (TMA) content, pH, and aerobic and anaerobic plate counts provided results that correlated well with each other and with SMO sensor results. A radial basis function (RBF) neural network was designed and used to classify the day of ®sh degradation (1±15) from the SMO sensor array response. # 2002 Elsevier Science B.V. All rights reserved. Keywords: Fish freshness; SMO; Sensor array; Food sensing
1. Introduction Fish is the most perishable of ¯esh foods, but is held in high regards for its ¯avor and health bene®ts [1]. On a pound for pound basis, seafood is as safe, if not safer, than other meat sources [2]. The safest seafood supply in the world is found in the US because the government, in conjunction with the food processing industry, has taken steps to ensure safety [2]. The largest dif®culty in maintaining safe and fresh seafood, according to the FDA, is that half of the seafood consumed in the US is imported from more than 35 countries and, therefore, does not undergo the same scrutiny as domestically produced ®sh, until it reaches the US [3]. Moist fresh ®sh will have almost no ®shy odor. The ®shy odor will develop with time after harvest, but should never be strong or objectionable. A strong ammonia smell is probably the result of protein breakdown and usually indicates old product, extended freezer storage, or possibly, mishandled product by storage at elevated temperature above 4 8C (temperature abused) [4]. In some species, the *
Corresponding author. Present address: 1 Idexx Drive, Westbrook, ME 04092, USA. Fax: 1-207-856-8728. E-mail address:
[email protected] (J. Hammond).
enzyme trimethylamine oxidase (TMAO) is responsible for protein changes and excessive moisture loss during thawing, producing a dry and/or coarse textured product that smells of ammonia [4]. Fish degradation after death is generally attributed to bacterial decay, enzymatic degradation, and lipid oxidation. Bacteria reside on the surface, gills, and in the gut of living ®sh, are the major cause of seafood spoilage. After harvest, the bacteria invade the ¯esh through the gills, skin, stomach lining, and blood vessels [5]. Enzymes, such as TMAO are present in some species, causing the ¯esh to soften, lowering quality, and producing more food for bacteria, accelerating spoilage rate. Oxygen in the air attacks oils and causes rancidity, off-odors, and -¯avors [5]. Decomposition does not always occur evenly within a single ®sh or between ®sh in a catch. Generally, decomposition begins in the anterior end of a ®sh and in the belly ¯aps, but exceptions have been observed. The rate and type of spoilage can vary with the time of year, species being harvested, and method of harvest [5]. When loss of quality occurs and spoilage begins, the process cannot be reversed and the product quality is lost. There are no universally recognized methods for determining the freshness of ®sh. There are many mechanisms for
0925-4005/02/$ ± see front matter # 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 4 0 0 5 ( 0 2 ) 0 0 0 1 1 - 4
114
J. Hammond et al. / Sensors and Actuators B 84 (2002) 113±122
®sh degradation, and the current best method of measuring the quality is by human sensory panels. A number of groups are attempting to come together and de®ne the problems and outline a systematic approach to creating an acceptable and accurate method for ®sh freshness monitoring [6]. Many groups are attempting to develop an ``electronic nose'' for ®sh freshness measurements, and most of them are developing sensors for speci®c gases that are assumed to be ®sh degradation products, such as trimethylamine (TMA) and dimethylamine (DMA) [7±10]. A successful sensor system for ®sh freshness will be trained with degrading ®sh. Atlantic salmon (Salmo salar), Atlantic haddock (Melanogrammus aegle®nus), and Atlantic cod (Gadus morhua) were selected for testing in this work. 2. Analytical techniques 2.1. Sensory evaluation There are many accepted methods for sensory evaluation of ®sh, but a standard method has yet to be de®ned [6]. A common technique is to train panel members to grade samples on a pre-de®ned linear scale [11]. The scale value then corresponds to a freshness value. For 2 months, a 15-member team was trained to quantitatively analyze the volatile odors of Atlantic salmon, haddock, and Atlantic cod. The goal of the training was to group all of the test people together in an agreed upon value for different ®sh samples. The training was broken down to sessions of 30 min or less (longer sessions are anti-productive because panelists nasal chemo receptors become desensitized). During the training sessions, the panelists were exposed to glass vials containing substances including DI water, TMA, salt water, ammonia, and ®sh samples including haddock, cod, salmon, and tilapia. Training continued until the panel members classi®ed samples to within a standard deviation of 2 on a 15-point scale. 2.2. Colorimetric technique for amine analysis A colorimetric method, commonly known as the modi®ed Dyer's method, was used to monitor amines in the ®sh samples [12]. The samples are mixed with acid and blended, and the solid portion is separated in a centrifuge. The supernatant is analyzed with a spectrophotometer against reference samples to determine the amine content. 2.3. Aerobic and anaerobic bacterial plate counts Bacterial plate counts are performed to show the total number of bacteria found in or on the ®sh muscle at various stages of degradation. The method used to perform the plate counts conforms to the ``Of®cial Methods of Analysis, Section 966.23'' of the Association of Of®cial Analytical Chemists (AOAC) [13].
Aerobic testing is repeated with media containing 1 and 10% tomato juice to determine lactic acid bacteria counts. The basis behind the tomato juice testing is to demonstrate the number of lactic acid bacteria in the total aerobic population. For anaerobic plate counts, a BBL GasPakTM Pouch System [14] is used to create an environment around the petri plate with 4% in
