New techniques for integral field spectroscopy - I : design, construction and testing of the GNIRS IFU.

We describe the integral field spectroscopy (IFS) capability of the Gemini Near-infrared Spectrograph (GNIRS) installed on the Gemini-South telescope. This makes use of the Advanced Image Slicer (AIS) optical concept of Content. Image slicing is, in principle, the most efficient technique of IFS. Our design is more compact and adaptable than the previous designs of this type so that the spectrograph can be switched to IFS mode simply by insertion of the integral field unit (IFU) into the focal plane. The near-optimal performance of the system makes it a good choice for instrumentation for future observatories, especially for the multiple-IFS systems required for extremely large telescopes (ELTs). The IFU produces good image quality and high throughput (>90 per cent at wavelengths ≥2.5 μm) which can actually exceed that of a spectrograph with a conventional slit of the same width, due to the use of anamorphism that reduces losses due to diffraction. This also results in a square spatial sampling element (spaxel) while simultaneously providing Nyquist sampling of the slit. At short wavelengths, the throughput is determined by the quality of the finish of the optical surfaces but exceeds 60 per cent at 1 μm. The three multifaceted mirror arrays in the slicing unit were made as monolithic units to avoid alignment errors between slices. This required the development of new freeform diamond-machining techniques which we describe. We present results on the testing of the optical components, and system-level tests to verify performance. We show that the IFU performs better than our expectations in terms of image quality and scattered light and that our model of the throughput, when fed with the results of our metrology of the surface quality, gives an accurate description of the performance measured in the laboratory. Paper 2 gives the results of on-sky testing to confirm the results of the laboratory tests and to verify performance by reference to existing data sets obtained with conventional techniques. The verification of our performance model is of great importance to the planning of future instrumentation of this type, as is the 'plug-and-play' nature of the integration with the spectrograph. Further details of the optical design and the optimization process are given in Paper 3