Spray drying is a widely used unit operation for producing particulate products directly from a liquid feed. Important processes that occur inside the spray dryer are droplet formation, droplet drying and interactions between droplets and recycled fines. Various studies have looked at the first two processes, but the latter phenomenon has received less attention. Literature on droplet-particle interaction which aims at quantitatively describing agglomeration in spray drying is scarce and mainly qualitative. For product quality the formation of agglomerates is often desirable. This thesis models and investigates the collisions of individual particles with single droplets. The surface tack of drying droplets has been identified as an important variable for the formation of agglomerates. In this thesis a novel method for measuring tack from the liquid phase has been further improved. The improvements are a more accurate load measurement, an automated control of the tack probe and an improved layout of the sample holder and probe. The key feature of the device is its ability to measure tack of drying droplets, whereas other devices measure tack by wetting a powder. Using our method the tack of a commonly spray dried product, yeast extract, has been measured. From these experiments it was found that with decreasing average moisture content the surface tack increases to a maximum. Below a critical average moisture content the surface of the droplet is dry and the tack rapidly decreases upon further drying. Another important parameter in determining the degree of agglomeration is the degree of penetration. If the particle penetrates the droplet too deeply, the agglomerate structure becomes too dense. To predict the penetration depth, a non-dimensional model has been developed. It describes the penetration of a particle into a liquid droplet during a head-on collision. It is based on a force balance and incorporates surface tension force, viscous force and capillary pressure force. The important parameters determining the collision outcome are the contact angle, the size of the droplet relative to the particle, the Reynolds and Weber numbers. For each contact angle an equilibrium penetration position exists, at this point the surface tension force vector is perpendicular to the penetration direction. Five different penetrations regimes are identified. At low Reynolds numbers, viscous forces dominate and the particle asymptotically travels towards the equilibrium position. Reducing the viscous drag force by increasing the Reynolds number results in initially overshooting the equilibrium position, but the surface tension force pulls the particle back, to attain the equilibrium in an oscillating motion. At even higher Reynolds numbers the particle fully penetrates the droplet, and reaches the centre of the droplet for even higher values for the Reynolds number. The ejection regime is found at high Reynolds number and low Weber numbers and the liquid should be non-wetting. Using the regime maps one is able to identify in which region a spray dryer is operating. Although the full penetration regimes are useful for capturing fines, it should be avoided when agglomeration is desired. The ejection regime should be avoided as well. To validate the model, impact experiments were carried out by dropping glass spheres on the surface of different liquids. These validation experiments were the first attempt to experimentally validate the collision of a single particle with a liquid surface. Besides yeast extract, which has non-Newtonian rheological properties, silicone oils with constant viscosities of 100 mPa•s and 1 Pa•s have been tested. The penetration over time for different impact velocities was determined by analysing high speed camera recordings. The typical penetration times ranged from 0.2 s to 2 s. To obtain accurate location data was recorded at frame rates up to 38 000 frames per second. Glass spheres, with a size of 2 mm were used to allow the visual tracking. Modelling the impacts showed that the model consistently predicted faster penetration times than were observed experimentally. The relative difference increased with increasing viscosity. A parameter fitting exercise showed that better agreement could be obtained by using a higher viscosity and a higher contact angle in the model. With this knowledge the most likely factor influencing the model-experiment mismatch was identified as being the dynamics of wetting of the particle surface. It was also found that using the dynamic contact angle in the model would improve its results. The non-Newtonian characteristics of the yeast extract resulted in the particle rebound and the formation of an air cavity upon impact. The tack measurement technique and penetration model presented in this thesis will be useful tools for the design of spray dryers. Recommendations are made for further model improvement. The experimental validation is the first attempt to validate the presented model. Future improvements are recommended and suggestions are presented.