ABSTRACT This project addressed two areas in the flavour industry – flavour reformulation required as a consequence of ingredient changes and flavour stability in solid food products. The aim of the flavour reformulation study was to evaluate both instrumental and modelling approaches to reformulate commercial flavourings for food products of different fat contents. A strawberry flavouring was chosen, which delivered an acceptable flavour in pectin jelly (0% fat) but did not perform so well in chewy candy (8% fat). Aroma release profiles in people eating these two sweets were measured by APCI-MS, the relative release differences for each flavour compound was determined and then used to reformulate the strawberry flavouring so its release in both sweets was similar. The sensory performance of candies confirmed a significant difference between the reformulated and original flavour (p < 0.05). The modelling approach was based on compound hydrophobicity (Log P) and the fat content of the food. This was used to estimate relative differences in flavour delivery for products with two varying fat contents. Flavourings were reformulated for yoghurt with reduce fat level, and the measured results showed good correlation with model’s prediction (r = 0.95). The aim of the flavour stability study was to explore the impact of product storage and flavour solvent in biscuits with vanilla flavouring. After eight weeks storage at 45 °C, up to 20% vanillin loss was measured, but sensory results suggested no difference in vanilla flavour perception. Texture analysis indicated that biscuits using triacetin (TA) as flavour solvent were more brittle than biscuits made with propylene glycol (PG). This was explained by X-ray CT analysis results which showed TA biscuits had larger pores than PG biscuits. Additionally, TA solvent provided higher vanillin stability during storage, so it should be a better choice than PG solvent used in flavourings for biscuits. IMPACT STATEMENT This research proved that it is feasible to apply laboratory-derived knowledge and adapt scientific techniques in a commercial context. It also demonstrated how the research findings can be transferred into the commercial field through two studies. The first study successfully illustrated two novel approaches to reformulate flavourings between food products of differing fat contents. Comparing with traditional trial and error process, this study demonstrated that instrumental and modelling approaches can be more effective ways for flavour reformulation in the flavour and food industry. The cost of flavour reformulation including the reformulation time and the cost in-use can therefore be reduced considerably. The results of this study have been presented to company’s major clients worldwide, and this has been applied commercially in a range of products including the successful reformulation of a £50 M (2010, brand value) soft drink into new formats (2011). The second study increased the understanding of the flavour stability within the stored food products. This ensured the flavour company to deliver high quality materials not only to their clients before food manufacturing, but also maintained its quality in the food products during storage for their end users. Comparing two flavour solvents – propylene glycol and triacetin, the results of this study suggested that triacetin is a better solvent as it offered better flavour stability in the stored biscuits. The findings can also be applied to reduce the cost in-use for vanilla flavouring in biscuit, which is especially beneficial for food and flavour company when high cost natural vanilla flavourings are required.

Few publications on oil bodies or oleosomes seem concerned about their quality (chemical and physical) ex-vivo. This work attempts to identify the main factors (processing and pre-processing) that affect the quality/integrity of sunflower seed oil bodies recovered through a wet-milling process. The physical state of seeds during wet milling had a significant impact on the quality of the oil body suspension. Pre-soaking for 6 hours before wet milling and multiple washing with alkaline buffer (0.1M sodium bicarbonate) was performed to isolate high quality oil body suspensions. It was evident from different physical measurements such as particle size, ζ-potential and light microscopy that pre-soaking had a positive influence on the quality of oil body suspensions with no significant signs of aggregation or coalescence. It was also observed that the resultant washed oil body suspensions were highly surface charged (-28.4 ± 1.2 mV) indicating very stable suspension phase behavior. Washing oil bodies not only removes non-integral, extraneous proteins (derived from the seed matrix) but enriches the lipid content including Tocopherol (α-tocopherol: 491.6 mg/kg of washed oil bodies compared with 252.6 mg/kg crude oil bodies). Changes in the composition of oil bodies after washing have been observed before, but this research also monitored the size of oil bodies after washing, and our results indicate that certain factors can shift the distribution of droplet size. It is believed that any change in average size of droplets indicate the presence of disrupted oil bodies whose surface chemistry has changed enough to compromise their integrity on washing. The retention of droplet size on washing may, therefore, be diagnostic for the recovery of intact oil bodies. An assessment of the integrity of oil bodies recovered from sunflower seeds after accelerated aging (5 months) was carried out. Free fatty acid was more pronounced in oil rather than oil bodies, this could be due to the elimination of some of the free acid bound to oil body during washing. Although some minor variation was observed during seed aging, however, the oil bodies remained stable in the final suspension. The results indicate that oil body membrane was extremely robust under extreme conditions and the integrity of oil bodies was preserved. In addition, oil bodies obtained in this study were resistant to oxidation due to the presence of naturally occurring antioxidants (including vitamin E) associated with them. The results indicate that the physical barrier of surface membrane protein (oelosin) protect oil bodies against pro-oxidants.

Beer, as manufactured, is not at chemical equilibrium and undergoes complex chemical reactions during storage which can lead to undesirable changes in the sensory characteristics of beer. Since beer contains more than 600 volatile compounds, beer flavour change is a complex field of research and the dependable prediction of shelf life remains a top research priority. Chapter 1 places the work described in this thesis into this context. Chapters 2 and 3 provide details of the experimental methods used and their development, respectively, to evaluate flavour stability. The methods used and developed include: (i) the determination of one of the major antioxidants present in beer, sulphur dioxide (SO2), via distillation; (ii) a solid phase micro extraction (SPME)-GC-MS method with on fibre derivatization as a reliable detection method for aldehydes related to off flavours perceived in aged beer; (iii) measurement of the oxidative stability using electron paramagnetic resonance (EPR) spectroscopy. Chapter 4 describes oxidative stability measurements via EPR spectroscopy for three different lager-style beer qualities pre- and post-filtration. The trial included a large scale Kieselguhr frame filter, and two pilot-scale membrane filtration system. The results illustrate how EPR spectroscopy is sensitive to metal ion pick-up from traditional filter media and oxidative stability measures were evaluated versus oxygen pick up during filtration and sulphur dioxide content of the beer samples. The effect of a brewhouse addition of gallotannins on the flavour stability of a lager-style beer was investigated in Chapter 5. Pilot scale (16 hl) and large scale production line (1500 hl) experiments with gallotannins additions in the Brewhouse were performed. The effects of the different additions were monitored at key points of the production process and through to the final beer. Despite showing significant improvements in the pilot scale wort samples, the results for the related packed beer samples did not show clear flavour stability improvements. For the production-scale trials, only a very limited effect of improved flavour stability could be observed. In Chapter 6 a factorial experimental design was adopted to probe the interactions between seven factors known to impact on beer flavour stability. Chemical additions were made to a bright beer prior to bottling, to vary the following factors: total in pack oxygen, SO2, total iron, iso-α-acid and α-acid content, (+)-catechin and glutathione. Increased SO2 concentrations had the largest impact across the entire design space, resulting in reduced radical formation, staling aldehyde concentrations and improved sensory scores. The impacts of increased TIPO levels were rather limited. In contrast, a significant impact regarding the oxidative stability could be observed for increasing Fe concentration, highlighting the significance of pro-oxidative effects of transition metals.

The bitterness of beer remains one of its most important flavour attribute. However, complexity surrounding bitterness perception alone represents a significant challenge in its understanding, in addition to other factors relating to production processes and raw materials used in making beer. The aim of this research was to better understand beer bitterness in terms of its intensity and quality, and how hopping technology (e.g. quantity and time of hop addition during the brewing process), hop aroma and variety may effect perceived bitterness intensity and quality. This project combined both analytical and sensorial analysis, and used statistical techniques to identify the key compounds driving bitterness perception. Analytical techniques were employed to quantify bitterness-contributing compounds and analytical bitterness units in beers alongside descriptive analysis of perceived sensory bitterness. Perceived sensory bitterness (both qualitative and quantitative) was assessed using a purpose developed list of well-defined bitterness attributes, to aid the characterisation of bitterness quality in beers. The impact of cross-modal flavour interactions on the perception of beer bitterness intensity and qualities was also investigated at varying analytical bitterness levels. Lastly, the effect of hop variety, and how this significant raw material affects perceived bitterness quality was also investigated using three distinctively different hop varieties (Hersbrucker, East Kent Goldings and Zeus). Results revealed a significant impact of the analytical profiles of beers, derived from the type of hop products and hopping technology adopted in beer production, on the perceived bitterness quality of beer. In general, beers high in hop acid compounds and polyphenols were, as expected, perceived to be higher in bitterness intensity. The results further showed that bitterness intensity also played a role in the perception of bitterness quality i.e. whether the beer was ‘harsh’, ‘lingering’ or ‘round’. However, beers high in these compounds were not always of ‘negative’ bitterness quality, and correspondingly lower amounts of these compounds did not always result in ‘positive’ bitterness qualities in beer. Results on the impact of cross-modal flavour interactions showed that hop aroma plays a crucial role in beer bitterness perception, and can modify the perception of bitterness intensity and quality as well as temporal profiles of bitterness in beer depending on analytical bitterness levels. The investigation into the impact of hop variety on perceived bitterness quality revealed hop-derived bitterness profiles relating to individual hop varieties, and further suggests that careful selection of the hop variety used for brewing is important for controlling perceived bitterness quality. This research provides a comprehensive interdisciplinary approach to understanding perceived beer bitterness and its associated qualities, with findings suggesting that besides bitterness intensity, the quality and overall impression of bitterness as perceived by consumers are significantly influenced by raw materials, production processes and other factors such as aroma which might be overlooked by brewers - but are likely to be significant for consumer preference and the success of a brand.

Very High Gravity fermentations are an increasingly attractive proposition within the brewing industry as a means of energy saving and optimising process efficiency. However, the use of very high gravity ( > 20°P) wort is associated with a range of biological stress factors. Ethanoic and osmotic stresses have been widely analysed along with oxidative stress in relation to propagation and early stage fermentation. The aim of this research was to investigate the impact of wort gravity on oxidative stress, and understand how this affects the plasma membrane at VHG. Finally the effect altered ergosterol content was analysed. The characteristics of Saccharomyces pastorianus strains were assessed for their ability to withstand the increased pressures of VHG (22°P) fermentations. VHG fermentations showed increase ethanol production at the expense of fermentation length, and increased ethanol production during VHG fermentations was offset by an increase in fermentation length. All fermentations were observed to accumulate ROS and increased antioxidant levels, with levels being furtherelevated in the VHG environment. Further analysis of S. pastorianus strains indicated that the levels of oxidative stress observed in fermentation had a negative effect on membrane fluidity and increased damage of the plasma membrane was observed. Analysis of ergosterol enriched yeast indicated that although fermentation rate was increased, a trade off with alcohol and biomass production was observed. Furthermore in response to oxidative stress, ergosterol enriched yeast showed reduced tolerance, decreased membrane fluidity and increased membrane damage. This work will give further insight into the response of lager yeast to oxidative stress present during VHG fermentations.

In the brewing industry it is standard practice to propagate a pure yeast culture and inoculate (pitch) it into the fermentation vessel. Once fermentation is complete, yeast is recovered and reused in subsequent fermentations (known as serial repitching) until a decline in performance occurs or the required number of successive fermentations has been conducted. Propagation is currently required to initiate the entire process again, which requires additional equipment, energy, water inputs and time. It has long been proposed that Active Dried Yeast (ADY) offers an alternative method of yeast supply. Adoption of this innovation by the brewing industry has been low because of perceived issues with the fermentation performance of ADY, the availability of strains and hygiene concerns. In the current study the fermentation performance of ADY has been assessed with respect to viability, genomic stability, membrane integrity, yeast growth, attenuation, uptake of wort nutrients and aspects of flavour development. ADY requires rehydration before use and it has been demonstrated that viability is impaired in these slurries, though the extent of viability loss was dependent on strain and rehydration conditions. The source of cell death is unclear. Mitochondrial and genomic DNA integrity was assessed using a variety of techniques and shown to be unaffected by dehydration and rehydration. In contrast membrane integrity was affected. Changes in membrane fluidity, sterol content and fitness to perform could be detected in ADY. Performance of ADY in fermentation was also impaired. A lag in cell growth, attenuation and sugar and amino acid uptake were noted. Diacetyl formation occurred more rapidly and end fermentation diacetyl levels were higher for ADY. These differences were not maintained during serial repitching. It is proposed that ADY could be utilised to replace freshly propagated yeast, but direct addition to fermenters would require an improvement of performance during the first fermentation.

To reduce the amount of sodium chloride in breakfast cereals without changing their properties, it is necessary to understand the role of this salt. Hence, a model system was developed. This model, composed of native waxy maize starch, glucose and a mixture of amino-acids generated similar colour and residual volatiles after heating compared to commercial breakfast cereals. Systematically designed experiments used this model to study the influence of NaCl concentration (0 % to 5.44 %) on colour, residual volatiles and acrylamide formation. It was found that higher NaCl concentration led to darker products (p<0.05) and significantly decreased acrylamide formation in the model systems. However, it did not have a significant impact on residual volatile levels. These findings were confirmed by observations made on wheat, wheat and rice mixture, com and rice commercial cereals. The impact of NaCl on colour and acrylamide formation indicated that this salt might influence Maillard and/or caramelisation rates and pathways. As NaCl is a plasticiser, it can allow the rubbery state to be maintained for a longer period during heating, improving reactants' mobility and Maillard reactions. However, it was found using the model systems mixed with several types of plasticisers (NaCl, KCl or trehalose), that the NaCl plasticising effect was not the major influence. The models also demonstrated that the hygroscopic behaviour of NaCl was not linked to its impact on colour and acrylamide formation. In investigating salt's influence on starch, native waxy maize, cassava or potato starch were mixed with NaCl (0 to 4 %; moisture adjusted to 20 %) and were heated at 230°C for 45 min. Microscopic observations, wide angle X-ray, viscosity, intrinsic viscosity and DSC data all suggested that starch was degraded by the heat treatment, and NaCl accelerated starch break down into smaller molecules, i.e. glucose. The glucose potentially formed could then caramelise, which might explain the NaCl impact on colour formation in model systems and breakfast cereals. Among other tested salts, CaCl[subscript]2 and MgC1[subscript]2 also enhanced starch degradation during a heat treatment. In studying glucose solutions containing salt (NaCl, CaCl[subscript]2 or MgCl[subscript]2) and heated between 180 and 230 °C, it was observed that salt enhanced colour formation via caramelisation (p<0.05). Mixtures of glucose, amino-acids and salt (NaCl, CaCl[susbscript] or MgCl[subscript]2), heated under the same conditions, showed that salt significantly decreased colour formation (p<0.05), which was most likely generated via Maillard reactions. Hence, salts could slow down Maillard reactions, explaining why lower acrylamide contents were found in model systems and cereal products when NaCl was present. As NaCl seems to influence Maillard and caramelisation reactions, decreasing or removing NaCl from breakfast cereal recipes might not only alter the salty taste but also the overall flavour. CaCl[subscript]2 and MgCl[subscript]2 seemed to have similar or even more impact on colour formation compared to NaCl. Adding these salts to breakfast cereal products with a low NaCl content was found to compensate for the colour loss. Adding CaCl[subscript]2 or MgCl[subscript]2 also improved the overall flavour of breakfast cereals, even though it did not compensate entirely for the taste loss.

The differences between conventionally processed and extruded puffed rice were examined, with a view to determining the reasons for their different storage behaviour in confectionery. Differences between the two forms of puffed rice have been identified, both in their performance and their properties. Under certain viewing conditions, the starch granules in conventionally processed and extruded rice appear to be different. The starch granules in conventionally puffed Rice remain intact, but with no crystalline structure (i.e. no Maltese crosses are visible, though starch ghosts are abundant.) Cell walls of conventionally processed rice appear to be a layer of these gelatinised intact granules. The ghostlike granules have not undergone notable swelling, remaining slightly greater than 10μm in diameter, similar to the size of unprocessed granules. In contrast, the starch granules in extruded puffed rice are rarely visible, in either their birefringent Maltese cross form, or as gelatinised but intact ghosts. By altering the extrusion conditions, the microscopic appearance of the product and its behaviour in water or water vapour becomes similar to that of conventionally processed rice. The extrusion parameters required for this similarity to conventionally processed rice fall within a window that is relatively narrow when compared to the ranges of variables available. Under more severe processing conditions the granules lose their integrity because shear and heat disrupt them. Under less severe conditions, the granules form clusters that do not gelatinise. The conditions at which this product is made cannot be interpolated or extrapolated from samples made under other conditions. However, at these extrusion parameters, the extrudate does not expand. It is necessary to expand the extrudate separately from the extrusion process. A novel method for the analysis of multi-peak force responses from compression of low moisture puffed cereal products was developed. This enabled crisp and crackly to be distinguished instrumentally. The results correlated well with sensory evaluation.

This research project was designed to investigate the chemical and physical changes in pork pies during storage. Lipid oxidation and moisture migration were found to be the parameters of most importance, with protein cross-linkaging and colour changes in the meat filling "'having' less of a deteriorative effect on pie shelf life. The extent of lipid oxidation and the development of rancidity (taste and odour) were found to be affected by cooking, the age of the meat and back fat in the meat filling and by their storage conditions (temperature and time). An increase in all or any of these parameters was found to increase the rate of rancidity development. On the other hand the rusk and the seasoning mix used in the meat filling of the pies, were found to reduce the rate of rancidity development. This antioxidant activity was investigated further. In the rusk, substances produced by the maillard browning reaction during the course of its manufacture were found to be antioxidative. However prolonged heating, although it produced a rusk with high antioxidant activity, had lowered its water binding capacity sufficiently to prevent its use in the meat products where water has to be bound to an inert filler. The antioxidant component of the seasoning mix was found to be white pepper, with the ethanol soluble fraction exhibiting this activity. Lipid analyses showed that rancidity developed parallel to the loss of the unsaturated fatty acids which had a chain length of 16 or above carbon atoms; and a concomitant rise in the concentration of short chain fatty acids and aldehydes, especially those having five and six carbon atoms in the chain length. The level of hexanoic acid (C6: 0) rose as rancidity developed, being development most marked in the pies with the fastest rate of rancidity development (i. e. those devoid of both the rusk and the seasoning mix). Moistture migration, both from. the atmosphere and from the jelly into the pastry (especially into the brown outer layer of pastry) resulted in an increase in moisture content, and a decrease in texture, of the pastry. Threshold levels for these two parameters were established above and below which respectively the pastry was deemed unacceptable, and these values were used subsequently in the investigation to assess the effect of various products on shelf life extension. Simultaneous lowering of the relative humidity (r. h. ) of the external atmosphere, and lowering of the jelly aw to levels at which to prevent moisture migration into the pastry resulted in a 35 day shelf life for the pies. However the use of glycerol to reduce the jelly aw to 0.56 resulted in an unpleasant taste to the pie. Lowering the jelly aw to 0.84 gave an extension (of at least 2 days) to the usual 8 days shelf life. Unfortunately attempts to achieve this aw (0.84) by the use of binders and pfi, were unsuccessful. Similarly the use of cetyl alcohol as a moisture barrier within the pork pie was not effective. The formation of a gelatin-lactein cogel, showed excellent potential for reducing moisture migration in pork pies.

The Rice Krispies™ process consists essentially of the cooking of short or medium rice grains, followed by a mechanical compression between two rolls (bumping), a tempering step and a toasting operation (puffing) which expands the grains into the finished product. The objectives of this project were to clarify which molecular phenomena take place inside the rice grains during the process and to facilitate the improvement and optimisation of the process parameters. The composition and gelatinisation behaviour of seven different raw rice varieties used either in the United States or the United Kingdom were studied. The glass transition of the cooked rice material was determined by dynamic mechanical thermal analysis (DMTA), which allowed each step of the process to be mapped onto a temperature / moisture content state diagram. The bumping step of the process and its effect on the various molecular entities present in the cooked rice were studied in details. Amylose was found to form complexes with the lipids present in the material during cooking, and these reinforced the cooked rice grain structure. One of the roles of the bumping could be to partially dislocate these amylose-lipid complexes to enable the rice grains expansion at the puffing step of the process. The bumping also reduced the molecular weight of amylopectin, which is thought to weaken the rice grains structure and improve the puffed grains expansion. The mill gap between the bumping rolls was the key factor determining the quality of the puffed product. The temperature at which the rice grains are bumped had a secondary influence on the quality of the puffed rice, within the range 26°C to 57°C. The RVA provided a very effective method to categorise post-bumped samples and could be a useful diagnostic tool in case of production problems. Evidence of physical ageing and moisture content equilibration during the tempering step was found. Physical ageing did not have a detectable effect on the expansion of the rice grains, while it was important that moisture content was equilibrated inside the grains to ensure a proper puffing. Finally, correlations were found between the puffed rice grains expansion and the post-bumped grains height recovery, which made it possible to predict the quality of the expanded grains from the characterisation of the post-bumped rice. Overall, the project is a good example of the application of general starch science concepts to the study of a specific industrial process.