Food structure is essential both for the consumer perception of dairy products and for the producer to produce high quality products. A desired structure in a dairy product can be obtained with the right combination of raw materials, unit operations and added ingredients. Interactions between proteins, lipids, carbohydrates, minerals and water and how they are affected by process conditions are difficult to predict. Optimization of the process combined with the knowledge of structure formation could also contribute to production of foods at a lower cost e.g. by binding more water without compromising the sensory quality of low fat products or increasing yield. Understanding how a desired structure is formed demands for a better comprehension of microstructure.

In this study, unit operations were in focus, with special emphasis on a combination of microfiltration and ultra-high pressure homogenisation (UHPH). The microfiltration will result in a milk fraction more or less depleted from whey protein, and could probably in combination with UHPH treatment contribute to milk fractions and cheeses with novel micro and macrostructures. These novel fractions could be used as new ingredients to improve structure of the final dairy products. UHPH treatment provides the possibility of inducing structural changes in proteins by a purely physical and mechanical method without addition of enzymes or other ingredients. Throughout the studies of this thesis different methods of analysis have been used to quantify differences in milk matrices and other methods have been applied to try and understand the effect of unit operations on a microstructural level. Low-field nuclear magnetic resonance (LF-NMR), small-angle X-ray scattering (SAXS) and small-angle neutron scattering (SANS) were used as non-destructive methods for this purpose.

A significant changed structure was observed in skim milk depleted or partly depleted for whey protein, acidified and UHPH treated. Some of the properties of the UHPH treated skim milk depleted from whey protein observed in this study support the idea, that UHPH treatment has more pronounced effect with higher ratios of casein-to-protein. These samples had a higher apparent viscosity (more structure) and smaller amount of expelled serum (bind more water).

From an industrial point of view, it was of interest, if observations of increased water binding (and structure) in the concentrate could be transferred to a traditional cheese making process. A significant higher yield was measured in cheeses made from UHPH treated concentrate compared to samples produced without UHPH treatment.

The main part of the applied methods was able to predict difference in milk matrices produced with different unit operations and parameters. However, not all methods could separate products based on all applied changes and some of the methods are only applicable for liquid samples. Measurements
with SAXS and DLS were the only applied methods capable of separating products independent of treatment, even though the effects were less clear in model cast cheese using SAXS.

SAXS and SANS being non-destructive methods are valuable to obtain and gain information about microstructure in combination with destructive test. Working together with physicists to model the data could further parameterize the findings with a raised knowledge level about microstructure in dairy applications as a result.

Due to the limited difference in scattering length density between protein and serum phase using SAXS, it is not expected to be able to observe protein fragments in the serum phase, while using SANS a higher difference in scattering length density between protein and serum phase is calculated and the possibility to observe protein fragments in the serum phase could be possible. Results obtained in this study indicate, that the presence of protein fragments could be predicted by SANS, but further modelling is needed to verify this.

LF-NMR relaxation were utilised to obtain information about the water mobility (relaxation time), in diluted skim milk systems depleted from whey protein. Obtained results indicate that measuring relaxation times with LF-NMR could be difficult to utilize, since no clear relationship between relaxation times and expelled serum was found. However, it was possible to differentiate samples with
different ratios of casein and whey protein and differentiate between native and denaturated whey protein. The sensitivity of the method to differentiate between whey proteins in different states native or denaturated could be used in other dairy applications with higher content of whey proteins.

The use of UHPH treatment seems to be a promising unit operation to change structure and bind water in dairy based products. The UHPH treatment could be applied to partial parts of the milk and concentrate and contribute to a total better food experience of e.g. low fat dairy products. Furthermore the UHPH treatment could limit the need for use of additives in this kind of products. In
production of traditional cheese the studies have shown, that increased yield can be obtained, by the use of UHPH treatment.