22 Continuous-flow heat processing N. J. Heppell, Oxford Brookes University 22.1 Introduction: definition of the process The continuous-flow heat process is a thermal heat-hold-cool process where the foodstuff to be treated is pumped in continuous flow through heat exchanger systems where it is heated to a desired temperature, held at that temperature for a pre-determined time, then cooled to around ambient temperature. After heat treatment, the product is then packaged in an appropriate manner. This process is different from in-container processes, such as canning or retorting, in which the product is firstly packaged and sealed, and then heat treated. Different thermal processes can be applied, from thermisation and pasteurisa- tion through to full sterilisation depending on the temperature and holding time employed, and packaging method selected. Terminology is difficult; for pas- teurisation, the process is often called High Temperature Short Time (HTST); for sterilisation, it may be called Ultra High Temperature (UHT) or aseptic pro- cessing. The advantages of continuous-flow heat processes over in-container processes are the much faster rates of heat transfer that can be attained, which means that higher process temperatures can be applied, usually around 140 to 150°C, as opposed to a maximum of around 125°C for in-container processes. In addition, the slow heating and cooling rates in in-container processes can incur considerable thermal damage to the product before the desired process has been attained. A comparison between the time-temperature profiles for the two processes is given in Fig. 22.1. Disadvantages of the process in comparison with in-container heat treatments include the fact that the process is less inherently safe than in-container process- ing as there are more points of potential contamination. Adequate sterilisation of the packaging material is essential; contamination may occur during the packag- Continuous-flow heat processing 463 ing operation and all parts of process equipment after the holding section must be sterilised before processing commences. In addition, the product must be capable of being pumped and it is therefore limited to liquids: however, the product may be purely liquid or liquid with comminuted solids or fibres (e.g. milks, creams, fruit juices, tomato purée) or a liquid which contains substantial solid particulates, such as soups, stews and cook-in sauces. 22.2 Principles of thermal processing 22.2.1 Thermal degradation kinetics Both the thermal death of microorganisms and thermal degradation of biochemi- cal components of food have been found to obey first order chemical reaction kinetics (with a few exceptions). At constant temperature, therefore, the reaction rate is given by [22.1] where N is number of organisms, C is concentration of biochemical, k is the reaction rate constant and t is time. In thermal sterilisation technology, equation 22.1 can be rewritten as: [22.2] where N and C are the number of microorganisms and concentration of bio- chemical respectively at time t, N 0 and C 0 are the initial number and concentra- tion and D is the decimal reduction time. The decimal reduction time is the time N N and C C t D t D 00 10 10== -- -= -= dN dt k.N and dC dt k.C In-container sterilisation UHT sterilisation 140 0 0 30Time (min) Product temperature ( ∞ C) Fig. 22.1 Temperature profiles of in-container and continuous-flow heat (UHT) processes. 464 The nutrition handbook for food processors taken to reduce the number of microorganisms or concentration of biochemical by a factor of 10 (i.e. to a value 1/10 th of that initially). This decimal reduction time is constant, i.e. for any time interval D, there will be a reduction to one-tenth. When quantifying the effect of temperature change on the D value, there are two major models used: the traditional ‘Canning’ (constant-z) model and the Arrhenius model. The former is: [22.3] where D 1 and D 2 are the decimal reduction times at q 1 and q 2 respectively. The z value is the temperature change required to change the decimal reduction time by a factor of 10. The Arrhenius equation is: [22.4] where A is a constant, the frequency factor, E a is the activation energy, R is the gas constant and q k is the absolute temperature. These two models are actually mutually exclusive but will agree within experi- mental error over a relatively short temperature range, which is usually the case for death of microorganisms. For larger temperature ranges, which is more usual for biochemical degradations, the Arrhenius relationship has a better theoretical basis and is generally used. Research work which reports thermal death of different strains of micro- organisms or degradation of biochemical components usually gives D and z values at a defined reference temperature (often 121.1°C) or the activation energy E a and frequency factor, A. There are tables of data given in Holdsworth (1992, 1997) and Karmas and Harris (1988) for microorganisms and biochemical components. 22.2.2 Effect of change in temperature An examination of the kinetics for microbial death and for degradation of bio- chemical components shows that the z values for the former are in the region of 10°C, while for the latter they are around 30°C. This difference is the basis of UHT processing: by increasing the temperature of a foodstuff, the microbial death rate increases much faster than biochemical degradation and, for equal levels of sterilisation, higher temperatures will give a better nutritional and organoleptic quality food than lower temperatures. One major disadvantage of this is that some enzymes may survive, especially heat-resistant proteases and lipases. It is impor- tant, therefore, that as high a process temperature be attained as is feasible and this is usually in the region of 137°C to 147°C. 22.3 Process equipment and product quality 22.3.1 Selection of heat exchangers The selection of heat exchangers for continuous flow heat processes is dependent on the physical properties of the food to be processed, especially its viscosity, k A.e ERak = -q D D 1 z 2 10 21 = -()qq Continuous-flow heat processing 465 the presence of solid particulates or fibres and any tendency of the foodstuff to ‘burn-on’. Heating may be indirect, where the heating medium is kept separate from the product, or direct, where the heat transfer medium is mixed with the food product (almost exclusively steam). For indirect heat exchangers, the heating medium may be steam or pressurised hot water and cooling may be by mains water, refrig- erated brine or liquid refrigerant as part of a refrigeration cycle. The process plant is based on corrugated plate, plain or corrugated tube or scraped-surface heat exchangers, a description of which, as well as applications and relative advan- tages and disadvantages, is given in many standard food engineering texts. Where possible, a heat recovery section (often called a regeneration section) is incorporated into the process to minimise its energy requirements. In this way, heat recovery of up to 80–90% of the total requirement may be achieved. For detailed process arrangements, see Lewis and Heppell (2000). The rate of heating and cooling of the food product within the equipment is important, as can be seen later, and is maximised in this type of equipment in three ways: 1 By increased turbulence in the food, minimising the liquid-heating surface boundary layer and therefore increasing heat transfer. This is usually achieved by corrugating the heat transfer surface to form a convoluted channel con- figuration for the liquid to flow down. 2 By increasing the ratio of heating area to liquid hold-up in the equipment, i.e. by decreasing the size of the product channel. 3 By increasing the temperature difference between the food and the heating or cooling medium. Of these, the first two can be easily accommodated, but the last often cannot be used as many food products are heat-sensitive and will increasingly form a deposit on the heating surface, increasing heat resistance and reducing heat transfer. Direct heating takes two forms: 1 Steam injection, sometimes called steam-into-product, where steam is injected directly into the food through a steam injector nozzle and the steam bubbles condense. 2 Steam infusion, sometimes called product-into-steam, where the product is pumped into a pressurised steam chamber, forming a liquid curtain onto which the steam condenses. Direct heating is the most rapid heating method but suffers from the dis- advantages that the process is noisy and the steam must be of a culinary grade, using only permitted boiler feed water additives. In addition, during heating the condensed steam dilutes the product by adding about 10% extra water and can be handled in one of two ways: 1 The recipe for the feed to the process can be made more concentrated, so that after cooling (using a conventional indirect heat exchanger) the final product is at the correct concentration. 466 The nutrition handbook for food processors 2 An equivalent volume of water is removed from the final product to bring the concentration back to the original value, e.g. for milk, where adulteration is illegal. The latter may be achieved using a ‘flash-cooling’ process, where the foodstuff passes from the holding tube through a restriction into a cyclone under vacuum. The sudden decrease in pressure causes the excess water to vaporise (or ‘flash’) and simultaneously cools the product. By altering the level of vacuum in the cyclone, the same concentration of solids as in the inlet can be achieved, in the outlet stream. One disadvantage of this process, however, is that the large tem- perature drop in flash cooling means that heat recovery is much lower than for indirect-heating systems and the process is more expensive to operate. 22.3.2 Effect of rate of heating on product quality The rate of heating and cooling of the foodstuff as it passes through the process gives a measurable effect on the nutritional and organoleptic quality of the product. Direct-heating systems, both injection and infusion, give the fastest rate of heating and flash evaporation gives the fastest cooling rate; both are virtually instantaneous. For indirect heating systems, on the other hand, the rate of heating is controlled by several factors: 1 The temperature difference between the foodstuff and the heating medium. The difference is limited by the heat sensitivity of the foodstuff, especially its tendency to ‘burn-on’ to heated surfaces. 2 The area available for heat transfer, i.e. the area of contact between the food- stuff and the heating or cooling medium. One important factor is the pre- sence of solid particulates in the foodstuff. The channel size through the equipment must be a minimum of three times the particulate size, which reduces the ratio of heat transfer area to liquid volume in the process, severely reducing the rate of heat transfer. 3 The heat transfer coefficients either side of the heat exchanger wall. These are controlled by both the turbulence in the foodstuff or heating/cooling medium, and their thermal conductivities. In addition, the physical state of the heating medium (whether liquid or condensing steam) is important. 4 The heat recovery section. The greater the heat recovery, the cheaper the system is to operate but the larger physical size of process plant means the rate of heating is much slower. The time–temperature profile of a continuous-flow heat process can be deter- mined from physical measurements taken on the process. The temperature points are determined by sensing the temperature at key points in the process, e.g. at the inlet and outlet points of each heat exchanger and any holding sections. The resi- dence time in each section is determined by measuring the volume of process plant between these temperature points and calculating the time as: Continuous-flow heat processing 467 [22.5] From this, a time–temperature profile similar to that obtained for heat penetra- tion into a container can be constructed and used in the same way to calculate F 0 values, chemical changes etc. Typical time–temperature profiles for direct and indirect-heating processes are given in Fig. 22.2. For a further explanation of this area, the reader is directed to work by Reuter (1982) and Kessler and Horak (1981a, 1981b), who collected time–temperature profiles for a wide variety of milk UHT plants and calculated sterilisation and biochemical changes expected. 22.3.3 Effect of residence time distribution on product quality Not all elements of the fluid food will move through the equipment at the same rate; generally, the fluid near the wall will move more slowly and that in the centre of the flow channel will move more quickly than the average flowrate. This has implications for the sterilisation of the product in that, in the holding tube, the residence time at the process temperature is lower than expected and therefore either the product will not be sterile or the holding time must be increased to compensate. This must necessarily mean that all slower-moving foodstuff will be over-processed with concomitant thermal degradation. The greater the spread of residence times, the worse the loss of thermosensitive nutrients. The most important factor in the residence time distribution is the flow regime in the liquid; either mean residence time volume of section volumetric flowrate of product through it = Direct heating system Indirect heating system (low regeneration) Indirect heating system (high regeneration) Time (s) Product temperature ( ∞ C) Fig. 22.2 Time–temperature profiles for direct heating, indirect heating, with low regeneration and indirect heating with high regeneration processes. 468 The nutrition handbook for food processors ? Streamline flow (for slow flowrates or high viscosity fluids) where the fastest element of fluid has a velocity twice that of the average velocity. or ? Turbulent flow (for higher velocities or low viscosity fluids) where the fastest element of fluid is about 1.3 times faster than the average velocity. For further information relating to prediction and measurement of residence time distribution and its effect on sterilisation and quality, see Lewis and Heppell (2000). 22.4 Processing and key nutrients: proteins The impact of any continuous-flow process on nutritional components in a food will be similar in type to the appropriate in-container process. However, as explained earlier, the difference in z values between microbiological death and biochemical degradation means that the level of that impact will generally be substantially lower. Much of the detailed work on changes to nutrients during continuous-flow heat processing concerns milk and milk products, as research in this area has been on-going since the 1950s. As the technology is more recently being applied to other products, a large body of work on further foodstuffs is progressing and will continue to do so. In general, a variety of changes in proteins can occur at elevated temperatures, especially denaturation and its associated changes in properties such as the water- holding capacity, viscosity or whippability, development of flavours (especially sulphurous), aggregation or formation of precipitates. Proteins are known to have a maximum thermostability at their isoelectric points although at pH values on either side, thermostability decreases at different rates depending on the type of protein and its environment. The effect of inorganic salts may also be important; at low levels of salts, stability is generally lower but, at high levels, can either reduce or increase stability. Lysine and cysteine are degraded by heat; losses of the latter rarely exceed 25% during in-container processing and are generally negligible in UHT processing. In milk, the proteins are divided into two types, caseins (in micelle form) and whey proteins. The whey proteins (a-lactalbumin, b-lactoglobulin, bovine serum albumin and immunoglobulins) are in solution and start to denature at tempera- tures as low as 70–90°C. Typical denaturation levels in UHT are given for pas- teurised milk as 5–15%, direct heating systems as 50–75%, indirect heating systems 70–90% and in-bottle sterilised as 80–100% (Renner, 1979). However, denaturation has little or no effect on their nutritive value, biological value or true digestibility. Unfolding and denaturation of b-lactoglobulin is responsible for release of volatile sulphurous compounds and the subsequent flavour of UHT milk. Oxidisation of these compounds during storage, either by residual oxygen in the product or leakage of oxygen into the container, will reduce this cooked Continuous-flow heat processing 469 flavour in the first few days after production. Again, direct heating processes were found to give a lower level of volatile sulphurous compounds, partly due to removal of the compounds during flash-cooling and despite the lower residual level of oxygen also due to this. The casein micelles are relatively heat stable and only minor changes are found. However, this reaction is important, because denatured whey proteins aggregate onto the casein and interfere with coagulation by acid or rennet, giving a looser curd. The milk is not suitable for use in cheesemaking. Soy milk is increasing in popularity as a product in Europe and the USA but raw soybeans have been long known to have a poor nutritive value due to the high levels of trypsin inhibitors in them. It is important nutritionally to reduce these inhibitors to less than 10% of the original concentration, at which point they will not interfere with biological value of the protein. The conditions required to achieve this are quite severe, for instance the D value at 143°C is between 56 and 100s, depending on the pH. Thermal denaturation of the inhibitor has been reviewed by Kwok and Niranjan (1995). 22.5 Carbohydrates and fats For mono- and oligosaccharides, little direct degradation occurs at temperatures typical of UHT processing but there are several reactions that occur that may affect nutritional quality. Firstly, Maillard reactions may occur, depending on the composition of the food, i.e. the presence of reducing sugars and amino acids. This is covered in Chapter 11 and will not be further discussed here. It is inter- esting to note that one of the intermediate compounds formed during the reac- tions, 5¢-hydroxymethylfurfural (HMF), has been used to indicate the level of heat treatment received by milk. Secondly, another reaction of note in milk is the formation of lactulose, an epimer of lactose formed during heating, which has also been used to distinguish between levels of heat treatment. This compound has been used in European Union legislation to distinguish between pasteurised, UHT-sterilised and in- container sterilised milks, and can also be used to distinguish between direct- and indirect-heating processes. Although lactulose has a laxative effect at around 2.5 mg/kg, milks fortunately do not reach this level, being around 2 mg/kg for in-bottle sterilised and much lower for other milks. Native starches will gelatinise at relatively low temperatures (60 to 85°C) and have been found to break down under the high temperatures and high shear rates present in an UHT process, giving a product with a vastly reduced viscosity after heat treatment. To overcome this, it is necessary to use chemically-modified starches which have been specifically devised to be stable under these conditions. Rapaille (1995) found a highly crossbonded and hydroxypropylated waxy maize starch performed best in both plate and direct-heating UHT processes. Low cross- bonded starches, however, were found to foul the process plant rapidly and gave a high viscosity during the preheating stages of the process. 470 The nutrition handbook for food processors 22.5.1 Fats UHT processing has not been found to cause any physical or chemical changes to fats in milk and milk products. Milk is usually homogenised during treatment, and some instability of the milk fat globule may occur due to denaturation of the proteins in the milk fat globule membrane. Because of this, homogenisation after the heat treatment section is usually preferred, especially in direct-heating systems, but has the potential to cause recontamination of the product due to leakage through the homogeniser seals and general difficulty with cleaning this area. It is important to have a well designed homogeniser with aseptic design and steam seals on the piston seals. 22.6 Vitamins The heat stability of vitamins in foodstuffs during heating is extremely variable, depending on the foodstuff and conditions of heating, e.g. presence of oxygen. As vitamins often consist of different chemicals, all of which have vitamin activity but degrade at different rates, it is often difficult to measure vitamin loss in a mechanistic way; see the discussion on vitamin C below. Apart from straight- forward denaturation, reactions between some of the vitamins may also occur, giving further losses. The heat-sensitive vitamins are generally taken to be the fat-soluble vitamins; A (with oxygen present) D, E, and b-carotene (provitamin A); and some of the water-soluble vitamins, B 1 (thiamin), B 2 (riboflavin) (in acid environment), nico- tinic acid, pantothenic acid, biotin and vitamin C (ascorbic acid). Vitamin B 12 and folic acid are also heat labile but their destruction involves a complex series of reactions with each other. Vitamin B 6 is generally little affected by heat, but storage after heat treatment can cause high losses. Niacin and vitamin K are fairly stable to heat. As before, losses through degradation in continuous-flow heat processing will be similar to those encountered during in-container processing but at a lower level. In milk, vitamins A, D, E, pantothenic acid, nicotinic acid, biotin, riboflavin and niacin are all stable to heat (Burton, 1988). Thiamin (B 1 ), B 6, B 12 and vitamin C all degrade during sterilisation. Thiamin is the most heat labile of these and has been used as a chemical marker to define the UHT process for milk by Horak (1980) as the time–temperature combinations which produce a sterile product but have a loss of thiamin of less than 3%. The thermal degradation kinetics have been established and used to predict thiamin loss for different commercial UHT processes; again, thiamin was expected to have a better survival in direct heating processes than in indirect heating processes and, of the latter in processes with small heat recovery sections than those with large sections. Vitamin C loss, although generally in the region of 25%, is not significant because milk is such a poor source of the vitamin in the diet. This compares well to losses of 90% of the vitamin during in-container sterilisation. The loss of the vitamin is less simple than a degradation: vitamin C exists in two forms, ascor- bic acid and its oxidised form, dehydroascorbic acid. The former is relatively heat Continuous-flow heat processing 471 stable but the oxidised form is much more heat labile; losses of vitamin C are therefore related to the degree of oxidation of the vitamin rather than to the sever- ity of the heat process (Burton, 1988). This is known to apply to other products, especially fruit juices (Ryley and Kadja, 1994) and vegetables. There are some vitamin interactions which occur. In milk, vitamin B 12 and folic acid interact with vitamin C during heat treatment so that losses of these vitamins again are not a simple function of the degree of heat treatment. Folic acid is protected by ascorbic acid, the reduced form of vitamin C, which, if oxidised, will lead to higher losses of folate. Vitamin B 12 losses depend on the oxidative degradation of ascorbic acid, so these losses depend on the availability of oxygen and ascorbic acid, as well as on the thermal process. Although these changes are dependent on the degree of thermal process applied to the foodstuff, the remainder of the process must also be considered when evaluating the nutritional quality of the product. Changes during storage of the product over the usual three to six month shelf-life at ambient temperatures can be considerable, even though the container is designed to exclude light and oxygen. In addition, there is usually some pre-processing during manufacture of products; for instance, soups, stews and cook-in sauces are usually cooked or fried before sterilisation. Finally, handling of the product in the home is usually out of the manufacturer’s control and overheating or long standing time at warm temperatures can easily have a more severe effect on nutritional quality than any of the procedures described above. 22.7 Future trends A relatively new technology will only succeed commercially if it offers benefits to the processor, in terms of operating costs or other means of operational efficiency, or if it offers benefits to the consumer in terms of attributes that they are prepared to pay for. In-container processing using a metal can has a poor image and appears ‘old’ technology, although some development with plastic trays, pouches or bottles is in evidence. Continuous flow heat processing has the advan- tage over in-container processing of a fresher image, because the packaging is similar to that used for fresh chilled products. The food then has an improved organoleptic quality if processed correctly but is still ambient shelf stable. The question is whether these advantages will outweigh the disadvantages of increased technical difficulty and increased unit cost, mainly due to the high cost and low filling rates of the aseptic filler. The indications are that the technology will grad- ually displace the in-container process to a degree; the continuous flow thermal processing of low-viscosity liquid foods such as milk, fruit juices, teas and tomato products is already well established and more, generally innovative, food prod- ucts are being introduced, albeit at a fairly slow rate. This slow introduction is very useful and may even be deliberate; because the process has a high complex- ity a large body of processing experience needs to be built up and consumers are more likely to become accustomed to this type of product. 472 The nutrition handbook for food processors All difficulties in processing are multiplied when producing foodstuffs containing solid particulates. Some soups, stews and cooking sauces containing particulates up to 25 mm in size have already been developed, but the greatest potential of the process lies in this area. Technical problems still to be solved satisfactorily are: ? Pumping of the foodstuff up to approximately 500 kPa pressure at a relatively low flowrate, then releasing the pressure to atmospheric again aseptically after heating, when the solid particulates are soft and easily damaged. ? A knowledge of the residence time of the liquid and solid phases in a holding tube, to enable sufficiently long minimum holding time to be established. ? A knowledge of the heat transfer rate from the liquid (usually heated first) to the solid particulates as the solids are transported by the liquids through the heating and cooling sections and down the holding tube, to enable the solid to be sterilised throughout. ? The development of aseptic filling equipment capable of handling particu- lates, without them being trapped across the packaging seal and compromis- ing sterility of the container. There is much active research at the present into these problems and several innovative techniques have been devised to overcome or circumvent them. One technique of note is the use of time–temperature integrators to ensure that the product has been heat sterilised to the required degree. These are small particles containing some chemical or microorganism (a marker) with a known heat resistance, which may be put through the process (whether by itself or embedded at the centre of a solid particulate), recovered and the surviving marker evaluated. From a knowledge of the heat degradation kinetics of the marker, the heat treatment received can be evaluated and used to ensure sterility of the product (Maesmans et al, 1994). Research work into particle residence time distribution and liquid-solid heat transfer has been reviewed by Lewis and Heppell (2000), Lareo et al (1997) and Barigou et al (1998). Another approach is to circumvent some of the problems and to use alternative equipment. Impor- tant in this respect are ohmic heating (covered in Chapter 19) and single flow fraction specific thermal processing (FSTP), that is a method of using a device to hold back particles of different size ranges and ensure they have a minimum holding time. 22.8 Sources of further information and advice holdsworth s d (1992), Aseptic Processing and Packaging of Food Products, London and New York: Elsevier Applied Science karmas e and harris r s (1988), Nutritional Evaluation of Food Processing, New York: AVI Publishing lewis m j and heppell n j (2000), Continuous Thermal Processing of Foods, Gaithersburg: Aspen Publishers Continuous-flow heat processing 473 willhoftema(1993), Aseptic processing and packaging of particulate foods, Glasgow: Blackie Academic & Professional 22.9 References barigou m, mankad s and fryer p j (1998), ‘Heat transfer in two-phase solid-liquid food flows: a review’, Transactions of the Institution of Chemical Engineers 76(C), 3–29 burton h (1988), Ultra-high-temperature Processing of Milk and Milk Products, London: Elsevier Applied Science holdsworth s d (1992), Aseptic processing and packaging of food products, Barking, UK: Elsevier Science Publishers holdsworth s d (1997), Thermal processing of packaged foods, London: Blackie Academic & Professional horak p (1980), Uber die Reaktionskinetik der Sporenabtotung und chemischer Veranderungen bei der thermischen Haltbarmachung von Milch. Thesis, Technical University, Munich, Germany karmas e and harris r s (1988), Nutritional Evaluation of Food Processing, New York: AVI Publishing kessler h g and horak p (1981a), ‘Objective evaluation of UHT-milk-heating by standardization of bacteriological and chemical effects’, Milchwissenschaft 36(3), 129–33 kessler h g and horak p (1981b), ‘Testing and appraisal of Type 6500 ultra-high tem- perature heat treatment plant’, North European Dairy Journal 47(9), 252–63 kwok k c and niranjan k (1995), ‘Effect of thermal processing on soymilk’, Inter- national Journal of Food Science and Technology 30(3), 263–95 lareo c a, fryer p j and barigou m (1997), ‘The fluid mechanics of two-phase solid-liquid food flows: a review’, Transactions of the Institution of Chemical Engineers 75(C), 73–105 lewis m j and heppell n j (2000), Coutinuous Thermal Processing of foods, Gaithers- burg: Aspen Publishers maesmans g j, hendrickx m e, de cordt s v and tobback p (1994), ‘Feasibility of the use of a time–temperature integrator and a mathematical model to determine fluid-to- particle heat transfer coefficients’, Food Research International 27(1), 39–51 rapaille a (1995), ‘Use of starches in heat processed foods’, Food Technology Inter- national Europe, 73–6 renner e (1979), ‘Nutritional and biochemical characteristics of UHT milk’, Proceedings of the International Conference on UHT processing and Aseptic Packaging of Milk and Milk products, Raleigh NC Department of Food Science, North Carolina State University reuter h (1982), ‘UHT milk from the technological viewpoint’, Kieler Milchwirtschaftliche Forschungsberichte 34, 347–61 ryley j and kadja p (1994), ‘Vitamins in thermal processing’, Food Chemistry 49(2), 119–29