Instrument setup

The data are collected using a semiconductor chip called a charge-coupled device (CCD), which is a highly sensitive solid-state detector consisting of a two-dimensional array of light-sensitive elements. These light-sensitive elements generate images con- sisting of an array of picture elements (pixels) where each pixel corresponds to one

light-sensitive element. Each photon of light that falls on a pixel generates one electron- hole pair in the semiconductor, so the number of pairs generated depends upon the intensity of the radiation. As the electron is produced by a photon it is called a photo- electron. Once the exposure is completed the accumulated charge is converted into a digital signal. To do this the accumulated charge is read out as a tiny electric current that is amplified and converted into a number and expressed in analogue data units (ADU). This value does not represent the number of electrons detected in the pixel but it is proportional to it. An analogue-to-digital conversion (ADC) factor is applied to the ADU number to quantify the number of photons that were incident on the pixel during exposure. The ADC factor represents the number of photoelectrons per ADU. Table 2.1 lists the characteristics of the CCD used at the NTT.

Grisms, offset slits and filters

The Faint Object Spectrograph and Camera (EFOSC2) is used with Volume-phased Holographic (VPH) grisms. A grism is a combination of a prism and a grating arranged in such a way that light at a chosen wavelength can pass straight through so that light is dispersed but not deviated. Unlike classical gratings, which have a surface structure, VPH grisms disperse light by Bragg diffraction from the refractive index modulations within a thin layer of processed dichromated gelatin sandwiched between two glass substrates. Light is diffracted at angles corresponding to the classical grating equation but the diffraction must also obey the Bragg condition. Maximum efficiency is achieved when the incident angle and wavelength match the Bragg condition and is much higher than the efficiencies achieved with surface gratings. Any given VPH grism will diffract different wavelengths as it is tilted with respect to the incident beam. If a camera can be tilted with twice the tilt of the grism the spectrograph can be tuned to a range of wavelengths and resolutions with the same grism, thus making it a very efficient and versatile piece of equipment (Hill et al. 2003). For these observations EFOSC2 was used with grism #20 and the 0.3′′

blue offset slit. The offset slit has a fixed offset of 15mm and it is used to extend the wavelength covered by the VPH grisms. The

Table 2.1: Characteristics of the CCD used at the NTT

General characteristics

Type – Loral/Lesser, Thinned, AR coated, UV flooded, MPP chip

CCD size – _{2060 × 2060}

Image Size – _{2048 × 2048}

Pixel Size – _{15 microns × 15 microns; 0.157arcsec × 0.157arcsec} Field Size – _{5.2arcmin × 5.2arcmin}

Full well capacity – 104,000 electrons/pixel Dark Current – 7 electrons/pixel/hour Digital saturation – 65535 ADU

In fast readout mode

Bias – 201–210 (ADU)

Readout Noise – 12.6 (electrons)

Gain – 1.31–1.38 (electron/ADU)

CCD readout time – 24 s (1x1 binning)

wavelength shift resulting from using this offset slit is found using the equation: dλ = 45.5D × dY

where dλ is the wavelength shift in Angstr¨om, D is the dispersion of the grism in Angstr¨om per pixel (0.55 for grism #20) and dY is the displacement of the slit in mm (15mm for the 0.3′′

blue offset slit). Hence, the wavelength shift we achieved was 375.375 ˚A which altered the wavelength range to 5672–6772 ˚A. This combination of grism and slit gave a spectral resolution of ∆λ = 1.2 ˚A with a resolving power (per 2.2- pixel resolution element) of R ∼ 5500, where R = λ/∆λ. This combination of grism and offset slit covers the wavelength range of 5672–6772 ˚A, which includes the major λ5780, 5797, 5850, 6196, 6203, 6269, 6283 & 6614 ˚A DIBs. Following the recommen- dations in the EFOCS2 user manual, because a 0.3′′

slit was used a fast readout with 1 × 1 binning was selected to ensure sufficient sampling in the dispersion direction and to reduce the overhead time per target. To maximise the signal-to-noise ratio (S/N) we achieved and the number of targets that could be observed during the night we aimed to expose the target until just below saturation level. This was optimal because our targets were bright and had short exposure time, thus, the overhead times including readout for each exposure was a major contribution to the total observation time spent per target.

The slit was originally positioned at column 1100 on the CCD to avoid known bad pixels on CCD#40. However, grism #20 introduces a lateral shift of the beam, therefore, the slit was later re-positioned at column 1680 to ensure the photons arrived approximately at column 1107 which corresponds to the area on CCD#40 that is free from bad pixels. The V-band magnitudes of the target stars ranged from 2.9 to 8.4 with exposure times ranging from 3.4 seconds to 847 seconds respectively. The V Bessel #641 and Hbe Cont #743 filters were chosen for the acquisition images with the latter being a narrowband pass filter to use with the brightest stars (≤ 3rd magnitude) to avoid saturation in the shortest exposures. The V Bessel #641 filter has a central wavelength of 5470 ˚A with a Full Width Half Maximum (FWHM) of 1134 ˚A, The Hbe Cont #743 filter has a central wavelength of 4770 ˚A with a FWHM of 72 ˚A, The

observation blocks for each target were prepared with the Phase 2 Proposal Preparation Tool (P2PP); version 2.13 was used in 2011 and version 3 was used in 2012.

Exposure times

Exposure times were calculated using the EFOSC2 Exposure Time Calculator on the ESO website and amended to take account of the limitations in slit width and grism that were available to choose from. This was necessary because the VHP grisms and offset slits were new additions and the online calculator had not been updated to include them. In the Instrumental Setup part of this calculator the smallest slit width that could be chosen was 0.5′′

and grism #20 was only available from 2012. For the 2011 exposure times the Grism information table on page 11 of the EFOSC2 manual (Monaco & Snodgrass 2008) was used as a guide to estimate the exposure times for our set up. As the seeing would likely alter during the night and the targets were bright it was expected that real time changes to the exposure time would be needed. The information in the table stated that for grism#18 the dispersion was 1.0 ˚A per pixel and the resolution at FWHM was 8.19 ˚A per arcsecond. For grism #20 the dispersion was 0.55 ˚A per pixel and the resolution at FWHM was 8.19 ˚A per 0.5 arcsecond. The online exposure calculator was used for grism #18 and a 0.5′′

slit, the required S/N was set to be 2000. As the targets were bright, uncertainties due to the Poisson distribution were insignificant. Therefore, to make the exposure time amendments, the maximum intensity value returned by the online calculator was first divided by the square root of the resolution given for grism #18 (rounded up to the nearest integer), this value was then divided by approximately two thirds of the saturation level as given by the online calculator to give the number of exposures per target required to gain the S/N level required. To scale the exposure time that was given by the online calculator it was first multiplied by 0.55 to take account of the difference in the dispersion for grism #20 (as given in the manual) and then multiplied by five thirds to adjust for the slit used being 0.3′′

instead of 0.5′′

. So for the 2011 observing runs the exposure time were estimated as above and repeated 3 times for each target.

grism #20. Therefore, the only adjustment needed was to account for the 0.3′′

slit being used. To estimate the exposure times for this run the maximum intensity returned by the online calculator using grism #20 and a 0.5′′

was divided by the saturation level and rounded up to the nearest integer to give the number of exposures needed. The exposure times were adjusted by multiplying by five thirds as before. This lead to the estimate of needing 6 exposures of the target at the estimated exposure time.

As mentioned, the seeing conditions at the time lead to many live time amend- ments to these estimated exposure times. In poor seeing the photons were spread out more so further exposures or an increase in exposure time was needed. In exception- ally good seeing the image of the target star was very sharp and so often difficult to accurately place onto the slit to achieve optimum number of photons, in which case the observations sometimes needed the target to be reacquired on the slit and further exposures taken, whilst in such good conditions when the target was accurately placed the exposure times needed reducing to avoid saturation. The quality of the spectra were inspected as they were acquired; for targets where the maximum counts per pixel were < 20, 000 a further three exposures were made to achieve the required S/N. This was needed in the 2011 observing runs more than the 2012 observing run which justified the increase in the number or repeat exposures calculated for 2012.

In March 2011 the skies were clear with seeing varying between 0.4′′

and 2.0′′

. In August 2011 one and a half nights were lost due to bad weather conditions. For the remainder of the August 2011 run fast moving clouds were present and observations were made though cirrus and thicker cloud. Individual exposure times were altered to accommodate the fast changing conditions with seeing varying between 0.4′′

and 2.4′′

. In August 2012 cirrus and thicker cloud were present during most of the observing sessions with seeing varying between 1.0′′

and 1.8′′

. On the last night (August 21st) the wind was strong and northerly restricting the direction on which observations could be made with the dome closing due to strong wind at 05:20, loosing an hour and a half of observing time. In the afternoon before the start of each night a set each of bias frames, dome flat field (4 × 1, 000W quartz-halogen lamps) and arc frames for wavelength calibration (HeAr) were taken, this is described later.