Surface water tracing experiments help to identify the flow paths of in-stream mass transport. This provides useful information for developing mitigation measures for contamination, and thereby for solving environmental problems. Salts, fluorescent dyes and isotopes are commonly used tracer substances.
However, it is well known these tracers have several drawbacks, including, for example, a finite number of unique tracers with similar transport properties, which limits the possibility to repeat an experiment due to system memory. Additionally, it is difficult to perform multi-tracer experiments, due to a lack of specificity. Lastly, the detection limit becomes an important constraint when using a tracer in larger water bodies or over longer travel distances, due to the vast dilution. The recently developed silica encapsulated synthetic DNA nanoparticle with a superparamagnetic core (SiDNAMag), can theoretically overcome the aforementioned limitations of existing tracers. SiDNAMag particles can be recovered from a sample by magnetic separation. This limits the influence of water quality on the qPCR analysis and provides an easy
method for up-concentration of samples. However, with the presence of natural colloids, particulate matter,
and river bed sediments, SiDNAMag particles are likely to undergo complex interaction processes in addition to
dispersion and advection. Moreover, little is known about the possible SiDNAMag tracer sinks during transport.
The focus of this work is to identify the transport behaviour of the SiDNAMag tracer in terms of breakthrough
curves (BTC) and hydrodynamic dispersion as well as to study the interactions with the natural environment,
specifically water quality and sediments. This was investigated by comparing SiDNAMag particles to NaCl,
a widely used solute tracer, and silica microparticles. To do so, open channel injection experiments using
SiDNAMag particles were performed in the laboratory in different river water types and with the presence of
sediments at the channel bottom. The resulting BTCs were interpreted with a 1D advection and dispersion
model with transient storage (OTIS), to determine the hydrodynamic dispersion coefficient. Mass recovery
was calculated by integrating the BTC.
The results show that SiDNAMag particles have a reproducible but different transport behaviour than
NaCl solutes. The SiDNAMag particles arrive, peak and return to background concentrations earlier than NaCl
solutes. Furthermore, the interpretation of the data with a 1D advection and dispersionmodel showed that the
dispersion coefficient for SiDNAMag was lower (0.34E-3m2/s – 0.88E-3m2/s) than that of NaCl (0.81E-3m2/s
– 2.31E-3m2/s). Literature on solutes and colloids in the aquatic environment supports the possible difference
in the transport behaviour of solutes and colloids. No relation between the transport behaviour and presence
of bottom-sediments was found. The mass recoveries for SiDNAMag experiments were between 36% and
131%, which could be partly explained by the bottom-sediments which detained between 0.16% and 16.94% of
the SiDNAMag mass during the experiments. No relationship was found between the three natural water types
and the observed mass losses. Additionally, the silica microparticle injection experiment showed the same BTC
characteristics for silica microparticles as SiDNAMag though without mass loss, indicating mass loss occurred
in the lab analysis. Due to the uncertainties in the lab analysis, no accuratemass recovery for the SiDNAMag
tracer could be calculated. Furthermore, the observed scatter in the measurement data could be explained
by the uncertainties in lab analysis as well as possible differences in transport behaviour of solutes and colloids.
In conclusion, the results of the laboratory experiments with the SiDNAMag particle show that it is a
promising hydrological tracer in surface water tracing experiments to trace colloids. Thereby it can be a
valuable tool to gain information on the movement ofmicroparticles, like microplastics and pathogens, in the
natural environment. The next step towards achieving this goal is performing a field experiment, for which
upscaling of the SiDNAMag tracer production is required.

Colloidal particle swarms can show settling velocities over a thousand times as fast as that of a single particle. The transport of colloidal particles is thus strongly affected by the formation and behaviour of these particle swarms, which could have important implications for the widespread use and prevalence of colloidal particles. Swarming effects may for example pose possibilities and limitations to the use of (colloidal) DNA tracers in various systems. Furthermore, contaminant transport in blue-green porous infrastructure may be greatly enhanced due to the rapid settling of swarms of colloidal contaminants, resulting in rapid spreading of contaminants into the environment. To better understand colloidal particle swarms in these systems, this thesis studied the effect of particle concentration on the behaviour of colloidal swarms in varying confining geometries. To this extent, colloidal particles were injected at various concentrations in a macro-model that was suspended in a stagnant water column. The macro-model consisted of three different artificial confining geometries: a smooth fracture, a rough fracture and a pore network. Released swarms were imaged to determine their velocity, width, travelled distance to bifurcation and particle leakage rates from the swarm. In the smooth and rough fracture, the settling velocity of colloidal particle swarms was modelled with the Hadamard & Rybczynski (HR) equation, using the particle leakage rate of the swarm. The average swarm velocity in the pore network was modelled with equations that described the Boycott effect. In both the smooth and rough fracture, the average velocity of the swarm showed a slight decreasing trend with increasing particle concentration. Presumably, this was caused by the linear relationship between particle concentration and width of the swarm. Consequently, wider swarms experienced lower velocities due to increased hydrodynamic drag forces imposed on the swarm. In the pore network, a strong linear relation was found between the average swarm velocity and the particle concentration. A larger mass of the swarm may have been the cause for this. The travelled distance before bifurcation took place did not show a clear relationship with particle concentration. However, in the performed experiments, swarms did bifurcate at a threshold width in both the smooth and rough fracture. This threshold width increased with concentration. The HR model did not fit the swarm velocity in the smooth fracture well, because confining forces of the fracture walls enhanced swarm velocity, which was not accounted for by the HR model. Yet, in the rough fracture the HR model did fit well to the observed swarm velocity since the increased drag forces of wall on the swarm counteracted the confining forces. The effect of particle concentration on the average swarm velocity in the pore network was sufficiently well modelled in terms of the Boycott effect. It was recommended that in future studies laminar flow is introduced to enable better interpretation of results for natural systems. In addition, a more simple pore network geometry provides a more gradual transition to the used complex pore network geometry. This would allow for clearer investigation of swarm behaviour in porous media. Also, more research on swarm behaviour after bifurcation may provide more insight in the longevity of swarm effects. To determine if colloidal particle swarms may occur in natural fractured media, results of this study should be compared to a field study. Both physical and chemical disturbances may severely hinder colloidal swarms. For both fractured and porous media, injection into a natural sample may provide more insight in behaviour in natural systems. It is however unlikely that swarming effects play a big role in urban colloidal contaminant transport in porous media.