PDS_VERSION_ID = PDS3 OBJECT = INSTRUMENT INSTRUMENT_HOST_ID = ESO INSTRUMENT_ID = IRSPEC OBJECT = INSTRUMENT_INFORMATION INSTRUMENT_NAME = "Infrared Spectrometer" INSTRUMENT_TYPE = "SPECTROGRAPH" INSTRUMENT_DESC = " Instrument Overview =================== IRSPEC is a cryogenically cooled grating spectrometer equipped with a 58 x 62 pixel SBRC InSb array with 76 micrometer pixel size. It covers the 1 micro- meter to 5 micrometer wavelength range at resolving powers of 1000 to 5000 with a nominal slit length of 2 arcmin. The instrument is attached permanently to one of the F/11 Nasmyth foci of the New Technology Telescope (NTT) and hence free of instrumental flexure effects. It employs an optical de-rotator in front of the slit to counter the field rotation at the telescope focus and to permit orientation of the slit at any position angle on the sky. IRSPEC is remotely controlled from the HP A900 'New Technology Instrument' (NTI) host computer via form filling and 'mouse' clicking on a menu bar. An on-line display of standard instrument and detector status parameters is available on a RAMTEK screen. Data acquisition is performed through the pre-processing system (PPS) which receives commands from the host. Images and 1-D traces can be displayed in real time on a monitor connected directly to the pre-processor. The display can be used for infrared 'peaking-up'. Brighter objects are centered using the NTT guide probes or the IRSPEC slit viewer. Final data are sent to the A900 where they are stored as individual 2-D IHAP files. IHAP is available on-line on the A900 with a RAMTEK monitor for image display and a graphics terminal for obtaining 1D spectrum plots, traces, etc. Optics ====== Light enters the spectrometer via the de-rotator which is installed between the NTT adapter A and IRSPEC. The de-rotator converts the NTT F/11 beam to F/8 at the slit. After passing through the entrance window, the F/8 beam is converted to F/7.4 by a field lens located immediately in front of the slit which also re-images the pupil onto a cold stop at the off-axis parabolic collimator. The latter directs the 10 cm diameter collimated beam towards one of the two 'back-to-back' mounted diffraction gratings (12 x 15 cm ruled area) operated in the Littrow mode. Grating # 1 (300 lines/mm) is used in the middle infrared and grating # 2 (600 lines/mm) in the near infrared. The F/2 Pfund type camera then focuses the spectrum on the detector array. The gratings are supported in a cradle mount which permits their interchange by a 180 degree rotation about an axis parallel to the ruled surface and orthogonal to the rulings. Both gratings have blaze angles approximately 37 degrees and can be rotated through approximately 25 degrees. The slit is a continuously variable moving blade system. Magnification between the detector and the slit is 3.7 so a 76 micrometer pixel corresponds to 280 micrometers on the slit or 2.2 arcsec. The slit blades are polished and slightly tilted to permit slit viewing with a TV camera. Behind the slit is an 8 position wheel containing the order sorting filters. The Littrow mount is off-axis by 10 degrees and the slit is tilted by 15 degrees relative to horizontal coordinates. This particular spectrometer arrangement causes a pronounced spectrum line tilt and curvature. The tilt angle omega(tilt) is given by omega(tilt) = arctan(sin gamma(0) x 2 tan omega) - theta (1) where sin omega = m x lambda / 2D x cos gamma(0) (2) with theta = 15 degrees, slit tilt gamma(0) = 10 degrees, off-axis angle of Littrow mount m grating order lambda center wavelength D grating constant Mounting ======== All the cooled optical functions are supported by, but thermally isolated from, a rigid optical bench which remains close to room temperature. The optical bench supports pass through stainless steel vacuum bellows attached to the vacuum vessel in order to decouple the latter from the spectrometer. Cooling of the optics to approximately 80 K is by means of silver straps or copper braids attached to the radiation shield which, in turn, is cooled by a continuous flow of liquid N2 (LN2) through a tube attached to its bottom plate. LN2 also passes through a heat exchanger sandwiched between the two gratings. The LN2 is supplied from an internal storage tank which is refilled twice a day. Temperature sensors automatically control the nitrogen flow to limit temperature gradients during cool down and to maintain the final steady state temperature. The detector array is cooled to 30K by a closed cycle cooler. Detectors ========= The detector at IRSPEC is a 58 x 62 pixel array from SBRC with 62 pixels aligned along dispersion direction. It is a hybrid type detector with backside illuminated InSb diodes bump-bonded to an X-Y addressed switched FET output multiplexer. The array has a quantum efficiency of 0.89 at 2.85 micrometers, 99.7% of pixels operable, and a well capacity of 1 x 10**6 e- . The array is operated in DC mode which is possible because of its high stability. The ADU to electron conversion factor is 85 e-/ADU. Dark Current and Noise Two major sources of dark current are present. The diffusion current, due to thermally generated charge carriers in the semiconductor, and the generation- recombination current due to emission of charge carriers through mid-band traps, add up to approximately 200 e-/sec at 30K. The internal back- ground of the instrument increases the dark current to approximately 400 e-/sec. Individual pixels exhibit noise which is due to Shot (Poisson) noise, Johnson noise, and 1/f noise. Shot noise is associated with statistical variations in the arrival rates of photons (i.e. the rate of generation of photo current) and in the generation and drift of dark current carriers across the semi- conductor. Johnson noise is thermodynamically fundamental to all resistances, including that of the detector. 1/f noise increases as the sampling frequency is lowered. Its origin is not understood. Read noise is defined as the RMS deviation of the signal read out of a pixel after zero integration time. Spatial noise, i.e. pixel to pixel variations present after flat fielding, determines the ultimate S/N that can be achieved. Data Modes ========== Two read out schemes are implemented for the SBRC array. Before an inte- gration, a reset switch is closed and the detector is set to an initial voltage Vi. This causes the detector and FET capacitances to be charged up. An integration is started by opening the reset switch. The effective capaci- tances are discharged by the photon and dark currents. In triple correlated read out, the detector voltage is sampled at reset (R1), at the beginning (S1), at the end (E1), and at second reset (R2) after the end of the inte- gration. The main contribution to the read noise results from the video line (kTC noise). It varies on time scales longer than typical detector integration times. This allows an interpolation of the noise during the integration; the signal voltage is thus obtained by Vs = (E1- R2) - (S1- R1). The read noise in triple correlated read out is approximately 680 e- RMS. The read noise is significantly reduced by non destructive read out. In this option, the output voltage is continuously sampled at a rate of 28 s**-1 without resetting the detector. The final signal voltage E1 is obtained by a least squares fit through the individual measurements. For detector inte- gration times around 10 sec, the combined dark and read noise is approximately 150 e- RMS. Figure 2.2 shows the read noise as a function of the detector integration time. It increases for integration times above one minute due to the internal background. Detector Response Pairs of columns are read out simultaneously via two independent output lines. A pronounced odd - even effect between adjacent columns appears in the detector response because the output lines have different gains. The difference is corrected to first order by software. There is an intrinsic non-linearity of the discharge of the total capacitance because both Idark and Cdet are functions of the bias voltage which changes during integration. The present software does not correct for the nonlinearity which is estimated to amount to 10% at maximum for this array (McCaughrean 1988). Acquisition and Storage The central unit is the NTI A900 computer on which the IRSPEC tasks are implemented. Data acquisition on the SBRC array is controlled through the pre-processor (PPS). Commands and replies from the A900 host system are sent via a simple RS232 connection. Final data are sent back to the A900 via a fast fiber link and stored as 2-D IHAP files. All IRSPEC commands are issued from the LU54 instrument console. Two RAMTEKs, LU38 and LU39, and a graphics terminal, LU72, are used to display final images, instrument and detector status parameters, and 1-D spectra. Processing ========== The pre-processing system is based on VME hardware. It contains a MC68030/25 MHz CPU board (Eltec E6) and a RAM board with 2 x 0.5 MBytes solid RAM disk cartridges. The software works under the OS-9 operating system. It handles the data acquisition, the host communication and the real-time display. On the PPS, the interface program CI waits for commands from the host, checks the correctness and triggers one or more programs depending on the command. After initialization of the acquisition hardware at start detector data are continuously read out upon an interrupt of the sequencer. The data are stored in an input-ring-buffer as pixel values. The mean and standard deviation are calculated and the result is stored into an output-ring-buffer. Individual measurements and final averaged data are sent on request to the NTI A900 host system. The PPS software contains a real-time graphics display task. It displays data frames, 1-D plots, and various integration parameters. The task has facil- ities like autocut (median filtered), display of pixel values for the various data frames, 'keeping images' for later display, subtract fixed pattern frame, etc. IRSPEC Software The IRSPEC software on the NTI-A900 consists of several programs: The obser- vation task IRSPC executes commands from the user or external programs. Its main functions are to set up the instrument and detector, to start exposures and transfer final data to tape. The spectrometer control task ICTRL receives commands related to the setting of spectrometer parameters sent by IRSPC. The detector control task IDTRL executes detector commands. Commands from the terminal are sent to the various destination programs via the Terminal Handler (TH). The display of messages on the terminal screen is handled by the Screen Handler (SH). It divides the terminal screen into two parts: in lines 1 to 5 replies from any destination program are displayed. Scrolling is disabled because of the need to write asynchronous replies. The instrument control task IRSPC communicates with the NTT computer via an Ethernet link. Both NTI and NTT computers keep the same, updated parameters in their respective memory (twin-pool). From IRSPC, any program, both on the NTI and NTT computers, can be accessed by typing its name. Calibrations ============ Flat Fielding In a procedure similar to that employed for optical CCDs, reasonable flat field corrections are obtained at wavelengths lambda < 2.5 micrometers using a halogen lamp. It is in general sufficient to obtain flat field exposures at the beginning and at the end of the night. To account for the non - linearity of the array, take flat fields at an exposure level similar to that of the objects. The halogen lamp is bright at wavelengths between 1 micrometer and 2.5 micrometers and the detector saturates for DITs larger than a few sec. Perform flat field measurements for the instrument settings used for the science exposures, including the various central wavelengths, grating number/orders, and filters. At wavelengths lambda > 2.5 micrometers, the thermal emission from the calibration unit provides a good flat field source. The use of standard stars as flat field sources is not recommended. Many of the stars previously believed to have flat continua in the near IR do in fact show stellar absorption features which, at the resolution of IRSPEC, have depths of a few percent. The odd - even difference in the response of the array is corrected to first order by software. An odd - even effect may still be present at a low level after flat fielding. If your observing program aims to measure spectra with high S/N, obtain science exposures with central wavelengths shifted by one or any other odd number of pixels. The ultimate S/N that can be achieved is limited by the spatial noise of the array. It cannot be improved by the accuracy with which flat field corrections are carried out nor by an increase of the total integration time. Whenever spatial noise is dominant, the faintest spectral features distinguishable are defined only by the uniformity of the array and the intensity of the background. Operational Considerations ========================== The position angle of the slit with respect to the sky is defined in the usual way: North = 0 degrees, angle measured positive to East. The angle is set by specifying the rotator (i.e. adapter A) offset angle (command PRSET> RTOF, see section C.1). The slit is conventionally set to a width of 2 pixels corresponding to 560 micrometers, i.e. 4.4 arcsec. This ensures an adequate sampling in lambda. The near infrared wavelength region accessible to ground based observations may be divided into two regimes: the 1 - 2.5 micrometer wavelength region and the thermal infrared at longer wavelengths. The near IR is dominated by non- thermal emission by the polar aurora, OH and O2 emission lines, and near IR nightglow. Vibrationally excited OH is produced in a reaction of Ozone with atomic hydrogen that takes place in a thin layer (approximately 10 km) between the mesosphere and the ionosphere (approximately 80 km height). The emission is highly variable on time scales of a few minutes. Pronounced diurnal intensity variations result from changes in the OH photodissociation rate which depends on the Doppler shift of the solar spectrum with respect to the predissociated OH absorption lines. Spectral variations result from changes in the rotational temperature of OH. Telluric O2 emission in the IR is limited to wavelengths around 1.27 micrometers and 1.58 micrometers. It arises from electronically excited O2 which is a photodissociation product of Ozone. The wavelength windows accessible from ground based observations are deter- mined by the atmospheric absorption lines of molecular species, mainly H2O and CO2. Figure B.1 contains the transmittance of the atmosphere between wavelengths of lambda = 1 - 5 micrometers. The edges of the atmospheric windows are highly variable. The photometric atlas of the solar spetrum (Delbouille et al. 1981) can be consulted if a detailed knowledge of the atmospheric transmittance is required. A proper flux calibration requires the observation of a standard star with the same instrument setting used for the science exposures. Hot stars have few absorption lines in the infrared, but can not be used if atomic hydrogen or helium features are investigated. For observations in bad atmospheric windows, choose a star as close as possible in air mass to the object in order to get an accurate flux calibration. If the detector gets saturated, a memory of the previous image is present on the array. To clean the array, perform a number of Dummy read outs (e.g. 5 Dummy read outs with DITs of 1 sec). The number of Dummy read outs is specified in the Exposure Parameters Form. Changing from one photometric band to the other results in a motion of the source along the slit by a few arcsec. The displacement can be determined by the observation of a standard star at similar wavelengths. As a major change compared to IRSPEC observations prior to the installation of the SBRC array, sky chopping and telescope beam switching have been removed from the observing menu. The detector is operated in DC mode, and FINALs up to several minutes on source. Observations in the J, H, and K Bands Analogous to chopping, cancellation of bright and variable sky emission is achieved through observations of a reference position close to the source. For compact sources, the sky observation may be obtained by observing the source at a different slit position. Total integration times per position are limited to approximately 10 min during the night when the sky emission is stable in order to obtain a proper sky line subtraction. Before and after sunset and sunrise, significant variations in the sky emission can occur at time scales below 5 min. Note Poor cancellation of sky lines may arise from significant spectral and intensity variations of the emission lines at times shorter than the inte- gration time. They may also result from small grating drifts after changing the wavelength. The stability of the grating is checked through the DITs stored as fixed pattern frames during integration. Variations in the sky emission or grating instabilities show up as deviations from a uniform, noisy pattern. Observations in the Thermal Infrared Sky lines are no longer dominant at wavelengths longer than 2.5 micrometers because of the large background from thermal radiation from the atmosphere. As shown in Table 3.6, the detector saturates at small DITs. Shot noise of the thermal background photons dominate rather than read noise. Because of this, the sensitivity is severely decreased. Sources without an optical counterpart are therefore searched and centered at 2 micrometers where the S/N is in most cases higher. The stability of the thermal emission compared to the frequency of sky observations is the limiting factor that determines the sensitivity that can be achieved. Sky observations should be carried out at intervals of typically one minute. The actually employed detector integration time is different from that specified in the Exposure Parameters Form. The correct DIT is displayed in the status display after the form is exited. It is also written into the header of the IHAP file stored on the NTI-A900. Because of the small DITs used in the thermal infrared, the actual integration time may differ significantly from the DIT specified. System Performance ================== The resolving power R of IRSPEC at the various combinations of grating number and order are listed in Table 3.8 (Melnick et al. 1989). The filters are selected if the parameter Auto Filter Selection in the IRSPEC Instrument Set Up Form is set to TRUE. Note that the resolving power listed in Table 3.8 is given for a slit width corresponding to two pixels. Higher resolving powers are possible for narrower slit widths. (A graphical representation of R as a function of wavelength is illustrated in Figure 3.1 of Melnick et al 1989.) Table 3.8 contains preliminary results of sensitivity measurements at the centres of the standard photometric bands. The listed limiting magnitudes and fluxes per pixel correspond to 1 sigma noises derived from observations of a standard star using 60 sec of integration time on the star and 60 sec on the sky. For point like sources, explicit sky observations are avoided by observing the source at two different slit positions. It is thus possible to gain a further factor of root 2 in S/N. Bibliography ============ [1 ] Wallander A.: 1993, Remote Control of the 3.5m NTT User Guide, ESO Operating Manual No 17. [2 ] D'Odorico, S., Ghigo, M., Ponz, D.: 1987, An atlas of the Thorium-Argon Spectrum for CASPEC in the 3400-9000 Angstrom region, ESO Scientific Report No. 6 [3 ] Dekker. H., Delabre, B.: 1987, Applied Optics, 26, 8, 1375 [4 ] Dekker, H., Delabre, B., D'Odorico, S.: 1986, SPIE, 627, 339 [5 ] Gilliotte, A.: 1992, Image Quality Filters Catalogue, Internal ESO publication [6 ] Melnick, J., Dekker, H., D'Odolico, S.: 1989, ESO Operating Manual #4 [7 ] Prieur, J.-L., Rupprecht, G.: 1990, Efficiencies of EMMI, ESO internal report [8 ] Wilson, R. N., Franza F., Noethe L., Andreoni G.: 1991, Journal of Modern Optics, 38, 219. " END_OBJECT = INSTRUMENT_INFORMATION OBJECT = INSTRUMENT_REFERENCE_INFO REFERENCE_KEY_ID = "N/A" END_OBJECT = INSTRUMENT_REFERENCE_INFO END_OBJECT = INSTRUMENT END