This thesis deals with light storage via electromagnetically induced transparency (EIT) in a rare-earth-ion-doped crystal. EIT is a quantum interference effect which enables precise manipulation of the linear and non-linear optical properties of quantum systems. Besides allowing cancellation of absorption EIT also permits the preparation of media in a state of highly anomalous dispersion. This led to the surprising effects of "slow" and "stored light" which are nowadays well established techniques in the field of optical quantum-state storage. The main goal of the work reported in this thesis is the implementation of an EIT-driven memory for coherent light pulses with enhanced storage capacity and extremely long storage durations.

The experiments were performed in a special kind of solid - a yttrium orthosilicate crystal doped with praseodymium ions (Pr³⁺:Y₂SiO₅). This medium offers the advantages of free atoms, i.e. spectrally narrow transitions and reduced interactions, while also providing high optical densities, good scalability and easy handling. Moreover, the absence of atomic diffusion in the solid medium allows spatially localized storage of large amounts of optical data.

In the first experiments to prolong the storage duration radio frequency magnetic pulses inverted varying phase evolutions of atomic coherences, which were initially prepared by the EIT light storage process. This extended the storage duration Δt from the pure dephasing time Tᵈᵉᵖʰ ≈ 10 µs to the coherence time T₂ ≈ 500 µs. These investigations revealed prominent oscillations of the storage efficiency versus the storage duration. The light storage process took place in nearly degenerate level systems. This led to constructive and destructive interference of the atomic coherences which were finally mapped on the restored light field. This "dark state beating" was observed for the first time in a solid. By application of weak magnetic fields the oscillations could be controlled and finally completely suppressed which led to a maximized efficiency for each storage duration.

Chapter 5 reports of investigations on three different techniques to extend the storage capacity of the EIT-driven memory. The storage of light pulses with imprinted image information increased the storage density compared to all former experiments on EIT-driven image storage by more than a factor of 40. The presented results represent the first implementation of this concept in a solid. Next, two techniques from classical holography were adapted to EIT-driven light storage. First, frequency multiplexing by EIT relies on simultaneous storage of light pulses in different ensembles of the inhomogeneously broadened medium. Second, angular multiplexing utilizes the phase matching of the applied laser fields to store information under different angles into the crystal. Both multiplexing techniques were implemented and combined with image storage. Moreover, a combination of both multiplexing techniques was demonstrated. This again increased the storage capacity of the EIT-driven memory.

The storage time in EIT-driven light storage is typically limited by decoherence. This obstacle can be overcome by application of external magnetic control-fields. For specific static magnetic fields, hyperfine transitions can be found whose transition frequency is - to first order - insensitive to external fluctuations. At such zero first order Zeeman shift (ZEFOZ) configurations, considerably longer coherence times of the relevant hyperfine-transitions can be observed. However, due to the Zeeman effect the external magnetic field also splits the levels and leads to a significantly more complex level structure. Hence, the medium was characterized by Raman heterodyne spectroscopy (RHS) with very high resolution. The optimal ZEFOZ-configuration was then determined by an automated search based on spin echo techniques. This procedure increased the coherence time from T₂ ≈ 500 µs to T₂ᵒᵖᵗ = 392 ms. The ZEFOZ-character of the determined transition was verified by means of RHS. Finally, by decoupling the relevant quantum system from residual noise the coherence time was further extended. This technique - known as dynamical decoupling (DD) - relies on a sufficiently fast driving of the system such that perturbations of the environment are averaged out. The obtained T₂-time of one minute represents the longest coherence time ever measured in Pr³⁺:Y₂SiO₅.

The last chapter discusses applications of these magnetic field techniques to the EIT light storage protocol. As the Zeeman effect splits all levels, a simple analytic preparation of the medium is not possible anymore. Hence a technique, originally developed in ultra-short pulse physics, based on self-learning pulse shaping was applied. An evolutionary algorithm (EA) imitated the principles of evolutionary biology to efficiently search for an optimized preparation sequence prior to a light storage experiment. In this work an EA was applied to such a complex, Zeeman-splitted and inhomogeneously broadened level structure for the first time. The EA found a solution, i.e. an optimized preparation pulse sequence which enabled light storage at ZEFOZ-conditions. The effect of the optimized pulse on the medium could be explained by comparison with numeric simulations.

The final step towards extremely long light storage times relied on dynamical decoupling via RF driving pulses. By application of 20000 π-pulses per second, the EIT-storage time could be extended above 40 seconds. This represents the longest ever observed storage duration in any EIT experiment and any medium. The storage duration approaches the ultimate time scale of the population lifetime of the relevant hyperfine states of T₁ ~ 100 s. To demonstrate the full potential of the presented concept, the experiment on long-term light storage was finally combined with image storage. Even after one minute the images could be retrieved with high contrast and no observable degradation of their structure. This outperforms all former experiments on EIT-driven image-storage with maximal storage durations in the range of 10 µs by more than six orders of magnitude.