Using the Information in this Chapter

Wavelength Ranges

The complete wavelength range, as well as the wavelength coverage per tilt for the scanned grating modes, is shown in a table and graphically. The exact wavelengths at the ends of the ranges for the MAMA detectors will depend on the MAMA monthly offsetting. This procedure shifts the spectrogram so that it falls on slightly different parts of the MAMA detectors from month to month in order to minimize charge depletion in the microchannel plates and can cause the loss of ±30 pixels from either end of the spectrogram in dispersion (AXIS1) and ±80 pixels in cross-dispersion (AXIS2). Spectral format plots of the STIS grating and central wavelength settings are available from the STIS web page under "Calibration." For the echelles, the monthly offsetting will project the extreme orders of formats off of the detector in some months. The wavelength ranges for the echelles given in the chapter therefore include only the orders which are guaranteed to project onto the detector. Whenever possible, choose a central wavelength that keeps your features of interest away from the extremes of the wavelength ranges.

Grating Sensitivities and Throughputs

This chapter contains plots and tables of sensitivities and throughputs for each grating mode. Determining Count Rates from Sensitivities in the Exposure-Time Calculation chapter explains how to use these sensitivities to calculate expected count rates from your source.

The total system1 spectroscopic point-source sensitivity, , has the unit:

counts sec-1 pix-1 per incident erg cm-2 sec-1 Å-1 for the MAMA and electrons sec-1 pix-1 per incident erg cm-2 sec-1 Å-1 for the CCD,

where:

• pix = a pixel in the dispersion direction.
• counts refer to the total counts from the point source integrated over the PSF in the direction perpendicular to the dispersion (along the slit).

Note that the spectroscopic point-source sensitivity does not include slit losses.

The spectroscopic diffuse source sensitivity, , has the unit:

counts sec-1 pix-1 pix_s-1 per incident erg sec-1 cm-2 Å-1 arcsec-2 for the MAMA and electrons sec-1 pix-1 pix_s-1 per incident erg sec-1 cm-2 Å-1 arcsec-2 for the CCD,

where:

• pix= a pixel in the dispersion direction.
• pix_s= a pixel in the spatial direction.

and are related through the relation:

where:

• m_s is the platescale in arcsec per pixel in the direction perpendicular to the dispersion.
• W is the width of the slit used in arcsec.

Here, we have assumed that the diffuse source has a uniform brightness over the area of interest and that the spectrum can be approximated as a continuum source. The throughput is defined as the end-to-end effective area divided by the geometric area of a filled, unobstructed 2.4 meter aperture (see Chapter 6).

The tables of sensitivities and throughputs given are derived for a point source in the largest clear aperture for each detector. The sensitivity plots give values for point and diffuse sources. In the plots in this chapter we show the diffuse source sensitivity for a 0.1 arcsec wide slit. For an extended continuum source, scales directly with slit width, as above.

For the echelles, the sensitivities given were derived from fits of a smooth curve as a function of wavelength to the measured sensitivities at the central wavelength of each order (i.e., they do not include the effect of the echelle ripple). The STIS Exposure-Time Calculator (see Chapter 6) will properly take the echelle ripple into account and should be used for more detailed S/N analysis.

Small but significant changes in the STIS sensitivity have been revealed by continuous observations of the same standard stars in the 52 x 2 slit. The largest changes are in the G140L, where the sensitivity depends on both time and temperature. Under normal operating conditions, the correction for the change in temperature is in the range of ±0.5%. No significant change with temperature was detected for the near-UV MAMA or CCD modes.

Following a correction for the correlation with temperature, the sensitivity of G140L decreases linearly with time. These losses vary with wavelength, ranging from a minimum of -0.8%/yr at 1325 Å to a maximum of -2.8%/yr at 1575 Å. For G230L, the mean sensitivity over all wavelengths increased by an average of 1.8% in the first 1.3 years and then began to drop. Two straight lines with different slopes fit the observed ranges. The changes before 1998.7 also vary with wavelength, ranging from a minimum of +0.8%/yr at 2550 Å to a maximum at 2250 Å of +2.1%/yr. The wavelength dependent slopes of the currently decreasing sensitivity range from zero at the short wavelength end of the band to -1.9%/yr in the 2300 - 2400 Å interval. Changes in the three low-dispersion CCD modes are less pronounced. For G230LB and G750L, most wavelength bins are consistent within a three-sigma uncertainty with constant sensitivity over time. Only one wavelength bin for G230LB and two for G750L show sensitivity losses with more than three-sigma uncertainty. Those changes are small and lie in the -0.2-0.4%/yr range. However, for G430L, the losses in every 200 Å wavelength bin beyond 3300 Å range from -0.5 to -0.7%/yr, with a high statistical significance of 2-9 sigma.

After the low-dispersion modes are corrected for sensitivity changes, the one-sigma photometric broadband repeatability ranges from 0.2% for G140L and G750L, to 0.4% for spectrophotometry in the other three modes. (See the STIS Instrument Science Report 99-07 for more details).

Less frequent monitoring of the M modes yields results generally consistent with those of the L modes, except that the mean sensitivities of G430M and G750M are declining at rates of about -1%/yr.

None of the changes described above is taken into account by the STIS exposure time calculator, which is only accurate to ~10%. Neither have they been incorporated into the STIS data-reduction pipeline (although they may be by Cycle 11). The sensitivities given in this chapter are derived from data taken in 1997 - 1998. Hence, they may be approximately corrected by the above annual percentage changes with respect to 1998.0.

Signal-To-Noise Plots

For each grating mode, a plot is provided to help you estimate the signal-to-noise (S/N) for a class of sources and a range of exposure times, corresponding to a fiducial taken at wavelengths near the peak of the responses. The fiducial wavelength is indicated in the ordinate label of each plot. To estimate signal-to-noise at alternate wavelengths, you can scale your source flux or magnitude by the relative sensitivities (or throughputs) at the wavelength of interest and at the fiducial. The point source plots show S/N as a function of F and of V+STMAG() for a range of exposure times; the diffuse source plots show I and V+STMAG() per arcsec² for a range of exposure times. Using STMAG units is natural in this plot given that a particular STMAG value corresponds to a flux distribution with a constant value of F. In producing these plots we assumed an average sky background (as described in Chapter 6) and the appropriate values for read noise and dark current for each detector. Note the following:

• The point-source S/N has been calculated per two dispersion pixels, and has been integrated over the PSF to contain 80% of the flux in the spatial direction.
• The point-source S/N calculations assume the use of the 52X0.2 slit for the first-order modes and the use of the 0.2X0.2 slit for the echelle modes.
• The diffuse-source S/N is calculated with the 0.2 arcsecond wide slit and the assumption that the slit is fully filled by a diffuse continuum source. The S/N is given per 2 spatial pixels and 4 spectral pixels for the CCD and is given per 2 spatial pixels and 6 spectral pixels for the MAMA (these are the equivalent spectral resolution elements for these observations).
• For the CCD, the plots assume CR-SPLIT=2 and CCDGAIN=1.
• The different symbols in the S/N figures delineate regions where different sources of noise dominate. A particular source of noise (readnoise for example) is presumed to dominate if it contributes more than half the total noise in the observations. Saturation is likewise indicated.
• The vertical dashed line indicates the MAMA bright-object observing limits; if F (or I) exceeds the indicated value, the observation would exceed the observing limits (recall that the observing limits are at slightly higher count rates than the screening limits given in Table 13.41). For diffuse sources we indicate only the local rate limit, since the global limit is dependent on the source extent. If no line is seen on the plot, it indicates the limit is higher than the range of fluxes plotted.

In situations requiring more detailed calculations (non-stellar spectra, extended sources, other sky background levels, etc.), the STIS Exposure-Time Calculator, located at the STIS WWW site under "Tools," should be used instead.

Follow these steps to use the S/N plots:

1. Look up in Table 13.1 the spectral type and wavelength region of interest of your target observation (e.g., G0 V @ 4300 Å). Interpolate in the table to get STMAG.
2. Add the V magnitude of the target to the interpolated value of STMAG derived from the table.
3. Find the appropriate plot for the desired grating, and locate V+STMAG on the horizontal axis. Read off the S/N for the desired exposure time, or vice-versa. Alternatively use F directly on the horizontal axis.
4. To get accurate values for repeated or CR-SPLIT exposures, one should use the sub-exposure time when consulting the plot, and then multiply the resulting S/N by , where N is the number of sub-exposures to be averaged. Recall that these plots assume CR-SPLIT=2 for CCD observations.

We now give a sample S/N calculation using these plots. Consider a V=18 star of spectral class B0 V, for which we want to derive the S/N for a 100 sec CR-SPLIT exposure in G430L with the CCD. We look up the B0 V spectral class and interpolate in the table between 4000 Å (STMAG = -1.20) and 4500 Å (STMAG= -0.78) to obtain STMAG ~-1 at 4300 Å. We thus have V+STMAG= 17. We look at Figure 13.12: and find this value on the horizontal axis. We locate exposure time 100 and find S/N ~10. This exposure is well below the saturation lines in the plot, so saturation is not a concern.

Plate Scales

In the grating information section, the plate scale (units: arcseconds/pixel) is given in the table for each grating. The values used have been obtained from imaging observations and have been approximated at 0.05 arcseconds/pixel for the CCD modes, and 0.025 arcseconds/pixel for the G140L and G230L MAMA modes (see "Image-Mode Geometric Distortion" on page 435). The other MAMA spectroscopy modes operate at a lower magnification, yielding a cross-dispersion plate scale of 0.029 arcseconds/pixel. Anamorphic magnification by the gratings further modifies the plate scales in the dispersion direction, particularly for the echelle modes. The relevant scales in both directions have been used in the generation of each grating's diffuse source sensitivity and signal-to-noise plots.

The exact level of anamorphic magnification is a function of grating and central wavelength. More detailed information on the CCD modes can be found in ISR 98-23 "Plate Scales, Anamorphic Magnification & Dispersion: CCD Modes" by C. Bowers and S. Baum.

Apertures

For each supported slit (aperture) we provide a table giving the name of the slit, length (in the spatial direction) and width (in the dispersion direction) of the slit as well as a table and plot of the relative throughput of the slit (with respect to a large clear aperture) as a function of wavelength. Recall, that the point source sensitivities that we have derived assume zero slit losses. Your calculations of exposure times must account for light losses for the desired slit. The aperture throughput information provided in the tables is based on models. Aperture throughputs measured in orbit are given in Instrument Science Reports 98-20 and 98-25.

Fiducials on Bars

Each STIS long slit has two fiducial bars, located 11.2 arcseconds above and below the slit center (see Figure 13.1 below). They have several purposes. First, the bars provide structural integrity for the long slits. Second, the image of the bars obtained in wavecal (and target acquisition) images is used by the calibration software to locate the projection of the aperture on the detector in post-observation data processing. Lastly, the bars can be used to occult a source thereby providing a coronographic spectroscopic capability for STIS. We continue to support use of the 0.5 arcsecond long bar on the 52X0.2 slit (the 52X0.2F1 aperture) for such observations. We refer you to Coronographic Spectroscopy for more information about performing coronographic spectroscopy, and Spectroscopic Mode Peculiarities for some caveats.

Figure 13.1: 52" Long Slits and Location of Fiducial Bars

Spatial Profiles

For each grating mode information about the cross dispersion (spatial) profiles is provided as follows:

• Plots of model fits to on-orbit data of the spectral profile in the cross dispersion (spatial) direction for a series of central wavelengths and gratings.
• A table of "encircled energies" containing two values: the percent of the total flux contained in the default calibration pipeline 1-D spectrum extraction aperture and the percent of the total flux contained in the central pixel.

Line-Spread Functions (Instrumental Profiles)

We show plots of predicted line spread functions (LSFs) for CCD and MAMA spectroscopic modes, as a function of wavelength and slit width. While these plots are based on prelaunch models of the PSFs at the aperture plane and detector PSFs from non-flight detectors, our comparison with on-orbit observations of internal lamps and external sources for the first-order modes indicates that the models are adequate for planning purposes. As part of the STIS calibration program, we will be making additional observations to characterize the LSF of the echelle modes and updates will be posted to the STIS web page in the "On-Orbit Performance" area as they become available.