Measurement Uncertainties in Fibre-coupled Spectrographs

The signal quality of fibre-coupled spectrographs can be limited by the inherent properties of the optical fibre. This is especially the case for applications that require high signal-to-noise performance and high spectral resolution. Examples include metallicity and age of star clusters, as well as investigations of Lyman-alpha absorbers. Extra-solar planet research in particular encounters its limitations due to the non-repeatability of the fibre response. Initially, a limited signal quality due to fibres seems counter-intuitive, since one of the most remarkable advantages of fibres is their signal stabilizing property, called image scrambling, which refers to the effect that the fibre output signal is largely insensitive to variations at the input side. However, the fibre photometric and barycentre response is sub ject to external parameters like stress, seeing and guiding variations. State-of-the-art instrumentation has attained a level of sensitivity where these effects will impact upon instrument performance, especially when advancing to a regime of spectral resolving powers where the quantized character of the standard optical fibre can be resolved, which manifests itself in modal noise. Unprecedented effort will be required in order to accomplish high resolving powers in the spectral and spatial domains with 40 m class telescopes. It is therefore essential to predict these fibre-related measurement uncertainties so that the performance of current and future instruments can be optimized. This thesis starts out with a phenomenological description of the different effects that give rise to fibre-related noise and its influence on the observables relevant to astrophysics, such as barycentre and photometric stability. Special emphasis is given to the photometric uncertainties related to modal noise, where first a theoretical model is outlined which in later chapters will be sub ject to experimental investigations. Subsequently, the barycentre repeatability due to incomplete scrambling is the subject of detailed investigation. The remaining sources of noise are estimated using experimental data as well as simulations and put in contrast with the other effects. Alongside the quantitative prediction of instrument instabilities, mitigation strategies will be presented and discussed. I conclude with a brief discussion of the impact of incomplete scrambling and modal noise on current instrumentation, the implications for future instrument pro jects as well as future work that will help to further understand and obviate the underlying mechanisms