Ground based near infrared adaptive optics as well as fringe tracking for coherent beam combination in optical interferometry required the development of high-speed sensors. Because of the high speed, a large analog bandwidth is needed. The short exposure times result in small signal levels which require noiseless detection. Both of these conflicting requirements cannot be met by state-of-the-art conventional CMOS technology of near infrared arrays as has been attempted previously[1][2]. The HgCdTe electron avalanche photo diode (eAPD) technology is the only way to overcome the limiting CMOS noise barrier of near infrared sensors. Therefore, ESO funded the development of the near infrared SAPHIRA 320x256 pixel e-APD arrays at LEONARDO [3][4][5][6][7]. SAPHIRA arrays have now become the devices of choice for control loops with unsurpassed performance [21]. This has also been demonstrated by the four wavefront sensors and the fringe tracker deployed in the VLTI instrument GRAVITY which set a new sensitivity standard in infrared interferometry [8][9]. It has also been demonstrated that APD arrays have extremely low dark current (1E-3 electrons/s/pixel) and may outperform conventional CMOS arrays for 100 second integrations when operated with moderate APD gains. For AO systems of extremely large telescopes and for co-phasing segmented mirror telescopes larger formats are needed. Therefore, a 512x512 pixel SAPHIRA array with 64 parallel video outputs optimized for pyramid wavefront sensing is in development. Since the SAPHIRA array has successfully passed radiation hardness testing it soon may be used for future instruments in space. Apart from the large array common voltage for high APD gain it can also be operated with voltages compatible with the space qualified SIDECAR ASIC [10].

A total of five near infrared SAPHIRA 320x256 pixel e-APD arrays have been deployed in the wavefront sensors and in the fringe tracker of the VLTI instrument GRAVITY. First results obtained by coherently combining the light of the four 8 meter telescopes of the VLTI have demonstrated diffraction limited performance with the CIAO adaptive optics system and a sensitivity improvement by a factor of 10 to 100 for interferometric observations. New APD developments since GRAVITY include the extension of the spectral sensitivity to the wavelength range from 0.8 um to 2.5 um. A larger format of 512x512 pixels is in development. At moderate APD gains the eAPD arrays have dark currents of <2E-3 e/s/pixel and outperform conventional state-of-the-art large format NIR science focal planes for integration times as long as 100s.

In 2007 ESO started a program at SELEX (now LEONARDO) to develop noiseless near infrared HgCdTe electron avalanche photodiode arrays (eAPD)[1][2][3]. This eAPD technology is only way to overcome the limiting CMOS noise barrier of near infrared sensors used for wavefront sensing and fringe tracking. After several development cycles of solid state engineering techniques which can be easily applied to the chosen growth technology of metal organic vapour phase epitaxy (MOVPE), the eAPD arrays have matured and resulted in the SAPHIRA arrays. They have a format of 320x256 pixels with a pitch of 24 μm. They now offer an unmatched combination of sub-electron read noise at millisecond frame readout rates. The first generation of SAPHIRA arrays were only sensitive in H and K-band. With the removal of a wide bandgap buffer layer the arrays are now sensitive from λ=0.8 μm to 2.5 μm with high quantum efficiency over the entire wavelength range. The high temperature anneal applied during the growth process produces material with superb cosmetic quality at an APD gain of over 600. The design of the SAPHIRA ROIC has also been revised and the new ME1000 ROIC has an optimized analogue chain and more flexible readout modes. The clock for the vertical shift register is now under external control. The advantage of this is that correlated-double-sampling and uncorrelated readout in the rolling shutter mode now have a duty cycle of 100% at the maximum frame rate. Furthermore, to reduce the readout noise rows can be read several times before and after row reset. Since the APD gain is sufficiently high that one photon produces many more electrons than the square root of kTC which is the charge uncertainty after reset, signals of one photon per exposure can be easily detected without the need for double correlated sampling. First results obtained with the fringe tracker in GRAVITY and the four SAPHIRA wavefront sensors installed in the CIAO adaptive optics systems of the four 8 meter telescopes of the VLTI have proven the unrivaled performance of the SAPHIRA eAPD technology. A future program is being assembled to develop eAPD arrays having a larger format of 1Kx1K capable of frame rates of 1.2 KHz. There are also good prospects to offer low dark current eAPD technology for large format science focal planes as well.

The only way to overcome the CMOS noise barrier of near infrared sensors used for wavefront sensing and fringe tracking is the amplification of the photoelectron signal inside the infrared pixel by means of the avalanche gain. In 2007 ESO started a program at Selex to develop near infrared electron avalanche photodiode arrays (eAPD) for wavefront sensing and fringe tracking. In a first step the cutoff wavelength was reduced from 4.5 micron to 2.5 micron in order to verify that the dark current scales with the bandgap and can be reduced to less than one electron/ms, the value required for wavefront sensing. The growth technology was liquid phase epitaxy (LPE) with annular diodes based on the loophole interconnect technology. The arrays required deep cooling to 40K to achieve acceptable cosmetic performance at high APD gain. The second step was to develop a multiplexer tailored to the specific application of the GRAVITY instrument wavefront sensors and the fringe tracker. The pixel format is 320x256 pixels. The array has 32 parallel video outputs which are arranged in such a way that the full multiplex advantage is available also for small subwindows. Nondestructive readout schemes with subpixel sampling are possible. This reduces the readout noise at high APD gain well below the subelectron level at frame rates of 1 KHz. The third step was the change of the growth technology from liquid phase epitaxy to metal organic vapour phase epitaxy (MOVPE). This growth technology allows the band structure and doping to be controlled on a 0.1μm scale and provides more flexibility for the design of diode structures. The bandgap can be varied for different layers of Hg(1-x)CdxTe. It is possible to make heterojunctions and apply solid state engineering techniques. The change to MOVPE resulted in a dramatic improvement in the cosmetic quality with 99.97 % operable pixels at an operating temperature of 85K. Currently this sensor is deployed in the 4 wavefront sensors and in the fringe tracker of the VLT instrument GRAVITY. Initial results will be presented. An outlook will be given on the potential of APD technology to be employed in large format near infrared science detectors. Several of the results presented here have also been shown to a different audience at the Scientific Detector Workshop in October 2013 in Florence but this paper has been updated with new results [1].

The performance of the current high speed near infrared HgCdTe sensors operating in fringe trackers, wavefront sensors and tip-tilt sensors is severely limited by the noise of the silicon readout interface circuit (ROIC), even if state-of-the- art CMOS designs are used. A major improvement can only be achieved by the amplification of the photoelectron signal directly at the point of absorption by means of avalanche gain inside the infrared pixel. Unlike silicon, HgCdTe offers noiseless avalanche gain. This has been verified with the LPE grown 320x256 pixel λc=2.5 μm HgCdTe eAPD arrays from SELEX both on a prototype ROIC called SWALLOW and on a newly developed ROIC, specifically designed for AO applications, called SAPHIRA. The novel features of the new SAPHIRA ROIC, which has 32 parallel video channels operating at 5 MHz, will be described, together with the new high speed NGC data acquisition system. Performance results will be discussed for both ROICs. The LPE material on the SWALLOW prototype was excellent and allowed operation at an APD gain as high as 33. Unfortunately, the LPE material of the first devices on the SAPHIRA ROIC suffers from problems which are now understood. However, due to the excellent performance of the SAPHIRA ROIC even with the limitations of present HgCdTe material, it is possible with simple double correlated sampling to detect test patterns with signal levels of 1 electron. An outlook will be given on further developments of heterojunctions grown by MOVPE, which eventually may replace eAPD arrays grown by LPE.

ESO has already published data from a preliminary laboratory analysis on the new mid-IR detector, AQUARIUS, at the previous SPIE conference of 2012, held in Amsterdam(2). This data analysis indicated that this new mid-IR Si: As IBC detector, from Raytheon Vision Systems, was an excellent astronomical detector when compared to previous generations of this detector type, specifically in terms of stability, read noise and cosmetic quality. Since that time, the detector has been deployed into the VISIR1 instrument at the VLT, with very mixed performance results, especially when used with the telescope secondary mirror, to chop between two areas of sky to do background subtraction and at the same time when many frames are co-added to improve the signal to noise performance. This is the typical mode of operation for a mid-IR instrument on a ground based telescope. Preliminary astronomical data analysis indicated that the new detector was a factor of two to three times less sensitive in terms of its signal to noise per unit time performance when directly compared to the old DRS detector that AQUARIUS was designed to replace. To determine the reason for this loss of sensitivity, the instrument was removed from the telescope and not offered to the ESO user community. A detector testing campaign was then initiated in our laboratory to determine the reasons for this loss of sensitivity, assuming that it was an issue with the new detector itself. This paper reports on our latest laboratory measurements to determine the reasons for this loss of sensitivity. We specifically report on indirect measurements made to measure the quantum efficiency of the detector, which can be difficult to measure directly. We also report on a little known source of noise, called Excess Low Frequency Noise (ELFN). Detailed analysis and testing has confirmed that this ELFN is the reason for the loss of instrument sensitivity. This has been proven by a re-commissioning phase at the telescope with the instrument and the detector. A new set of observing parameters and observational regime have been developed to help to mitigate the ELFN. We outline a possible explanation for the source of the EFLN, learnt from a literature search and discussion with the manufacturer.

The most promising way to overcome the CMOS noise barrier of infrared AO sensors is the amplification of the photoelectron signal directly at the point of absorption inside the infrared pixel by means of the avalanche gain. HgCdTe eAPD arrays with cut off wavelengths of λc ~2.64 μm produced by SELEX-Galileo have been evaluated at ESO. The arrays were hybridized to an existing non-optimized ROIC developed for laser gated imaging which has a format of 320×256 pixels and four parallel video outputs. The avalanche gain makes it possible to reduce the read noise to < 7 e rms. The dark current requirements of IR wavefront sensing are also met.

ESO has begun an ambitious mid-IR detector program with the funded development of a new Raytheon detector (AQUARIUS) and the further development of instruments to use 5 μm cut-off material Teledyne HAWAII-2RG detectors. Both these detector types are capable of extremely high readout speeds, through multiple readout ports, resulting in data rates in excess of 250 Mbytes/s. This has required further development of our new detector controller system (NGC) to allow it to operate at these very high pixel data rates. This has also entailed the development of new high speed pre-amplifiers which can operate at 60K to allow us to drive the long cable runs typical of an astronomical instrument. We report on the development and performance of our new higher speed NGC systems with particular regard to the operation of a Hawaii-2RG detector configured to use its high speed readout stages. We will present data on the performance of such at device, configured to operate in both slow and fast readout modes, with particular regard to noise versus pixel speed and also the optimization of the voltages.

Complementary to the overview of the optical detector projects, this paper lists the infrared projects that have been started or completed at ESO since the DfA2005 workshop [1]. New detector developments for the mid-infrared, the implementation of innovative near infrared avalanche photo diode array designs for AO wavefront sensing and interferometry and novel detector controllers on a single chip will be discussed as well. Reference will be made to the topics which are presented in more detail in other contributions to DfA2009 [2].

Teledyne Imaging Sensors (TIS) has developed a CMOS device known as the SIDECAR application-specific integrated circuit (ASIC). This single chip provides all the functionality of FPA drive electronics to operate visible and infrared imaging detectors with a fully digital interface. A Teledyne 2Kx2K silicon PIN diode array hybridized to a Hawaii-2RG multiplexer, the Hybrid Visible Silicon Imager (HyViSI) was read out with the ESO standard IR detector controller IRACE, which delivers detector limited performance. We have tested the H2RG HyViSI detector with the TIS SIDECAR ASIC in 32 channel readout mode at cryogenic temperatures. The SIDECAR has been evaluated down to 105 Kelvin operating temperature and performance results have been compared to those obtained with external electronics. Furthermore ESO has developed its own interface card to replace the JADE USB card provided by Teledyne. The ASIC controller is now being embedded in the ESO standard VLT hard and software environment. This paper provides an update on the recent development of the new ESO ASIC interface card. We find that the SIDECAR ASIC provides performance equal to external electronics.

The CALICO sensor is a pathfinder for the development of the future infrared high speed low noise detectors for AO. Low readout noise at high readout speed is accomplished by high gain and signal processing circuitry under each pixel. The high gain makes the detector very susceptible to instability if the system noise is too high. Lnpix3, the most promising structure, has a pixel gain of 400. In this paper we will report on test results and different measures we had to take getting the detector to work.

Recently ESO has commissioned the HAWK-I camera which is equipped with a 2×2 mosaic of λc~ 2.5 µm HAWAII-2RG arrays. The arrays have high quantum efficiency and achieve photon shot noise limited performance on the telescope. Using reference pixels it can be shown that the readout noise for most arrays is limited by the MBE grown HgCdTe material and not by the multiplexer or the data acquisition chain. Results obtained with the guide window of the HAWAII-2RG multiplexer will be presented. Inter-channel crosstalk and fringing in the detector substrate will be discussed. The dynamic range of detectors can be expanded by applying threshold limited integration (TLI) to the follow-up-the-ramp nondestructive sampling scheme. For substrate removed arrays a calibration technique based on the X-ray emission of Fe55 will be discussed.

Teledyne Imaging Sensors (TIS) has developed a new CMOS device known as the SIDECAR application-specific integrated circuit (ASIC). This single chip provides all the functionality of FPA drive electronics to operate visible and infrared imaging detectors with a fully digital interface. At the last SPIE conference we presented test and performance results of a Teledyne 2K×2K silicon PIN diode array hybridized to a Hawaii-2RG multiplexer, the Hybrid Visible Silicon Imager (HyViSI). This detector was read out with the ESO standard IR detector controller IRACE, which delivers detector limited performance. We have now tested the H2RG HyViSI detector with the new TIS SIDECAR ASIC in 32 channel readout mode at cryogenic temperatures. The SIDECAR has been evaluated down to 105 Kelvin operating temperature and performance results have been compared to those obtained with external electronics. We find that the SIDECAR ASIC provides performance equal to optimized external electronics.

Progress in FPGA technology has made it possible to build a very compact, versatile, low power detector array controller based on high speed serial link technology. All data and communication transfers run between back-end and front-end, and within the front-end over high speed serial links with transmission rates of 2.5 GBit/s. The serial-bus architecture offers high performance and system scalability for modular controller set-up.

Instruments for large 10-m class telescopes increasingly require high sensitivity large format focal planes. The high spatial resolution achieved with adaptive optics combined with multiple integral field units feeding high resolution spectrographs are driving the pixel performance and require large detector formats. In the infrared spectral range, the array formats have arrived at 2K×2Kpixels with both LPE and MBE grown HgCdTe on CdZnTe substrates. In the optical, fully depleted Si-PIN diodes of the same format are used. The light-sensitive diode arrays are hybridized to CMOS FET switched multiplexers such as the Hawaii-2RG array, which has recently been installed in one of the infrared instruments of the Very Large Telescope (VLT). Basic performance characteristics of the Hawaii-2RG arrays will be discussed such as the noise performance when a special technique of using reference pixels is employed. Larger focal planes are realized as mosaics of 2K×2K arrays. In order to increase the format of single arrays to 4K×4K and larger, the limited substrate sizes make it necessary to reduce the pixel size. However, with smaller pixels the coupling between pixels becomes a limiting factor for the detector point spread function. Fundamental calibration issues relevant to photon transfer techniques of modern CMOS active pixel sensors with special regard to the influence of interpixel coupling capacitances will be analyzed in detail. A novel technique will be presented to directly measure the point spread function generated by the capacitive coupling between adjacent pixels.

MUSE, the Multi Unit Spectral Explorer, is a 2nd generation instrument for the VLT. It is built by a consortium of European institutes and ESO. MUSE consists of 24 Integral Field Units each equipped with its own cryogenically cooled CCD head. The detector is 4096x4096 with 15 mu m pixels.In this paper we discuss CCD requirements and give an overview a the design status of the detector system. Due to the large number of units, each must be simplified to save cost and man power, and increase reliability. We present a novel preamplifier for use inside the compact detector head with pulse tube cooling. Adaptations of ESO's New General detector Controller (NGC) are presented.

The conversion gain of optical and infrared focal plane CMOS hybrid arrays is a fundamental parameter, whose value computes into the derivation of other parameters characterizing the performance of a detector. The widespread “noise squared versus signal” method used to obtain the conversion gain can overestimate the nodal capacitance of the detector pixel by more than 20% for infrared arrays and by more than 100% for Si-PIN diode arrays. This is because this method does not take account of the capacitive coupling between neighboring pixels. A simple technique has been developed to measure the nodal capacitance directly by comparing the voltage change of an external calibrated capacitor with the voltage change on the nodal capacitor of the detector pixel. The method is elaborated in detail and has been verified with a Si-PIN diode array hybridized to a Hawaii-2RG multiplexer using an Fe 55 X-ray source. It is also in good agreement with a stochastic method based on 2D autocorrelation.

This paper presents the design of a new ultra low background detector mosaic test facility for IR detectors with cutoff wavelengths up to 5.3 mu m and operating temperatures down to 20 Kelvin. In instruments with high spectral resolution such as ESO CRIRES, dark current and readout noise of the detector system is often the limiting performance factor. This test facility allows us to develop, optimize, test and assess the functionality and performance of IR detector mosaics for low flux applications prior to the integration in the instrument. Therefore detector tests must be performed under extremely low photon backgrounds of << 0.1 photons/sec/pixels. No cryostat window is used and the blackbody is contained inside the hermetically sealed radiation tight cold structure.

The New General detector Controller (NGC) at ESO aims at single-handedly covering optical, infrared and wavefront sensing applications. However, the detector requirements for optical, infrared and Adaptive Optics (AO) have increasingly diverged in the past years, presenting challenges in the design of the electronics. This paper extends the description of the general architecture of the controller [1] towards the specific front-end electronics to drive and read-out CCDs both for optical instruments and AO applications. In addition, the adaptation of the NGC front-end to the future ESO CCD head for AO and the control of the L3 devices are presented.