TECHNICAL REPORT Title: Consequences of Etalon Detuning for TFI Science Doc #: JWST-­‐STScI-­‐002190, SM-­‐12 Date: July 15, 2010 Rev: -­‐ Authors: A. R. Martel, A. Fullerton Release Date: 28 September 2011 Phone: 410-­‐ 338-­‐4888 1. Abstract We present a detailed analysis of the consequences of the position-dependent bandpasses (PDBs) that are inherent to optical systems that incorporate a Fabry-Pérot etalon, with particular emphasis on the Tunable Filter Imager (TFI). Simulations are generated for a variety of input spectral energy distributions: a flat spectrum, Vega, a planetary nebula (NGC 6543), and a high-redshift QSO. We show that even though the systematic change in wavelength across the field of view (“detuning”) is small, the corresponding variations in the integrated counts may be significant (up to ~25%) when the target is imaged near a wavelength characterized by a sharp spectral gradient. The detected flux can decrease or increase as a function of position depending on the interplay between: a) the shape of the source continuum across the bandpass, b) the location of a feature within the bandpass, c) the proximity of the bandpass to the edge of the blocking filter (which favors one wing of the Airy function over the other); and d) the change in the intrinsic shape of the bandpass. The PDB of the TFI has important implications for science observations. In particular, a judicious choice of dithers or mosaics may be necessary to mitigate these variations. Accurate photometry of astronomical sources may require substantially different corrections to be applied at different locations in the field of view. 2. Introduction Martel & Fullerton (2011) presented the basic operating principles of the TFI etalon and discussed the concept of wavelength “detuning”. Since the etalon in the TFI is not used in a telecentric configuration, light corresponding to different radial distances from the optical axis enters with a range of incident angles, !. This behavior imposes a positional dependence on the peak wavelength of the bandpass such that 2! !! ! ! = cos ! = !! cos ! ≈ !! 1 − , !eff 2 Operated by the Association of Universities for Research in Astronomy, Inc., for the National Aeronautics and Space Administration under Contract NAS5-03127 Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. (1) JWST-STScI-002190, SM-12 where d is the gap between the plates of the etalon; !eff is the “effective order”1; !! ≡ 2! !eff is the wavelength at the center of the field of view (i.e., where ! = 0°); and where the small-angle approximation for the value of cos ! has been used. As a result, even though an etalon is tuned to transmit a wavelength !! on-axis, the wavelength of the peak transmission is shifted systematically (“detuned”) toward bluer wavelengths in a circular pattern centered on the optical axis . For the TFI, 0° ≤ ! ≤ 4.2°, which reduces !! by only ! ! 2 ≈ 0.3%. In addition to the location of the transmission maximum, off-axis rays change the shape of the bandpass transmitted by the etalon; i.e., the Airy function. This dependence can be seen explicitly by combining Equations (34) and (3) of Martel & Fullerton (2011): !! 2!eff,! = 1+ ! ! ! sin2 2 2!eff,! = 1 + ! ! 2!" cos ! − !! sin2 ! ! !; !eff,! (2) !! (3) where !eff,λ is the “effective finesse” of the etalon and !! is the phase change upon reflection at each surface. Both these quantities may be wavelength dependent. The presence of the cos ! term causes the shape of the bandpass to depend on the distance of an object from the optical axis. In the following sections we analyze the consequences of this position-dependent bandpass (PDB) for the detected counts of typical astronomical sources when they are observed at different positions in the field-of-view (FOV) of the TFI. In particular, we present more realistic simulations generated for a variety of spectral energy distributions (SEDs) observed with the TFI. Our goal is to assess the impact of the continuum and strong spectral features on the variations of both the wavelength and the observed counts across the FOV for typical astronomical sources. 3. Simulations In the following section, we explore the consequences of “detuning” by examining its effect on the following SEDs, all of which are specified in the standard ‘flam’ units of erg/s/cm2/Å: a) A constant (“flat”) spectrum in “wavelength” units b) Vega c) The emission-line spectrum of a planetary nebula (PN) d) A high-redshift Ly! emitter 1 See Section 2.1 of Martel & Fullerton (2011) for the definition of “effective order.” Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -2- JWST-STScI-002190, SM-12 3.1 An Ideal Etalon with a Flat Spectrum 3.1.1 Peak Wavelength Our first simulation attempts to reproduce the expected behavior for the ideal case of a uniform (“flat”) input SED. In Fig. 1, Airy profiles are generated for the optical axis of an etalon with a constant reflectance and transmittance similar to those of the TFI: ! = 0.85 and ! = 0.15. The reflective finesse is therefore !! ≈ 20. A 180° phase shift is introduced to mimic the phase changes at the TFI plates. For an etalon gap of 4.0 µm, the peaks of the Airy profiles in the operational interference orders shortward (! = 3) and longward (! = 1) of the non-functional regions of the etalon occur at 2.0 µm and 4.0 µm, respectively. The input spectrum is a flat distribution normalized to unity and expressed in the standard ‘flam’ units of erg/s/cm2/Å. The blocking filters are also fixed to unity and are assumed to be symmetric about the peaks of the profiles, 1.8 – 2.2 µm and 3.4 – 4.6 µm. These simulations indicate that: 1. The Airy profile is always asymmetric in the sense that the red “half” of the profile contains more counts than its blue counterpart. 2. Detuning: As " increases from 0° to 4°, this asymmetry remains but the peak shifts to bluer wavelengths. To track the location of the peak, a 2nd -order polynomial was fit to the upper 20% of the output spectrum (magenta in Fig. 1) as a function of " . This wavelength detuning is shown in Fig. 2. We find that the ! for both the blue ! and ! red sides of the non-functional region of the TFI detuning etalon is an excellent match to Equation (1). In fact, all three lines (blue, red, and dashed) overlap perfectly in Fig. 2. The maximum detuning is ~0.3%, as ! expected. 3. PDB: As ! increases and the output spectrum shifts to the blue, the bandpass becomes slightly narrower; i.e., its resolving power increases. Since the counts are integrated under the profile over a fixed wavelength range defined by the blocking filter, a net decrease is expected. This is shown in Fig. 3. The maximum change is ~0.5% for the flat input spectrum of this simulation. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -3- JWST-STScI-002190, SM-12 Figure 1 Simulations for ! = 0° on the blue (top) and red (bottom) side of the TFI etalon’s non-functional region. The inputs are: Airy profiles (dashed black), flat input spectrum (green), and normalized blocking filter (cyan). The output spectrum (magenta = green x dashed x cyan) is shown in the bottom panels as well as the second-order fit to the peak. Constant input parameters typical for the TFI are assumed: ! = 0.85 !! ≈ 20 ; ! = 0.15; ! = 1; and an etalon gap of 4.0 !m. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -4- JWST-STScI-002190, SM-12 Figure 2 Wavelength detuning for the simulations described in Fig. 1 in absolute (top) and relative (bottom) wavelength. Blue: shift in the peak of the profile on the short-wavelength side of the non-functional region of the TFI. Red: same but for the long-wavelength side. Dashed: detuning described by Equation (1). The three lines overlap exactly in the bottom panel. Figure 3 Relative change in integrated counts due to PBD. Blue: output spectrum to the short wavelength side of the TFI non-functional region; red: same but to the long wavelength side. 3.1.2 Average Wavelength For simulations with SEDs of some astronomical sources, the peak of the output spectrum may not correspond to the peak of the bandpass provided by the etalon. Such cases arise, e.g., if the peak of the emission line is highly distorted or asymmetric (as for Ly! in a QSO, see §3.2.4) or if multiple emission lines fall within the bandpass, causing Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -5- JWST-STScI-002190, SM-12 the software to confuse the true location of the peak. For these spectra, we will instead estimate the centroid of the output spectrum (after multiplying with the Airy function and the blocking filter) with a flux-weighted average. By default, this average is calculated over the full range of the blue and red regions, 1.5 – 2.5 µm and 3.2 – 5.0 µm, respectively, and so may include some contribution from any “leaks” outside the blocking filter’s primary bandpass. An example is shown in Fig. 4 for the simulation of a flat input spectrum. Because of the asymmetry in the profiles, the centroid is always skewed to wavelengths slightly longer than the peak of the bandpass (top panel) and so the ideal situation described by Equation (1) is not expected to be recovered (bottom panel). Figure 4 Same as Fig. 2 except that the centroid of the output spectrum is calculated with a flux-weighted average. 3.2 The TFI Etalon In this section, we present simulations for the more realistic case of the TFI. The throughputs of the actual blocking filters (F158M, F175M, F200M, F229M, F246M, F349M, F425M, and F470M) are used. Except for the flat spectrum and Vega, the centroid of the output spectra is calculated as the flux-weighted average. To simplify the analysis and the interpretation, the effective finesse is assumed equal to the reflective finesse, i.e., defects such as surface irregularities on the etalon plates are ignored. The net effect of these effects is to degrade the resolving power of the filter; i.e., to increase the width of the bandpass (Martel & Fullerton 2011). Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -6- JWST-STScI-002190, SM-12 3.2.1 A Flat Spectrum In Figures 5 – 7, we show the results of a simulation with a flat distribution as the input SED. The etalon gap is finely tuned to 4.0073 µm so the peak of the output profile in the third order (short-wavelength side of the non-functional region) is located at 2.0 µm at ! = 0° for a more direct comparison with the results of §3.1.1. The corresponding peak in the first order (long-wavelength side) is at ~3.86 µm. As in §3.1.1, the detuning in wavelength is tracked with a 2nd-order fit to the peak of the output profile (magenta in Fig. 5). From Fig. 6, we find that the detuning in wavelength is slightly less severe than for the case of an ideal etalon and blocking filter. This is likely due to the structure of the blocking filter, which favors the red wing of the profile over the blue wing. In fact, for the TFI, it is impossible to achieve the ideal behavior described by Equation (1). On the other hand, the change in count-rate is more important for the TFI than for the ideal case, up to ~2% (Fig. 7) versus ~0.5% (Fig. 3). Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -7- JWST-STScI-002190, SM-12 Figure 5 Simulations for ! = 0° on the short-wavelength side of the TFI etalon’s non- functional region (top) and on the long-wavelength side (bottom). The inputs are: Airy profiles (dashed black), flat input spectrum (green), and non-normalized blocking filter (cyan). The output spectrum (magenta = green x dashed x cyan) is shown in the bottom panels as well as the 2nd-order fit to the peak. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -8- JWST-STScI-002190, SM-12 Figure 6 Flux-weighted wavelength shift for the simulations shown in Fig. 5 in absolute (top) and relative (bottom) wavelengths. Blue: shift of the profile on the short wavelength side of the non-functional region of the TFI. Red: same but for the long-wavelength side. Dashed: detuning described by Equation (1). Figure 7 Relative change in integrated counts due to PDB. Blue: output spectrum on the short-wavelength side of the TFI non-functional region; red: same but for the longwavelength side. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. -9- JWST-STScI-002190, SM-12 3.2.2 The Spectrum of Vega In these simulations, the input SED is the spectrum of Vega from the Kurucz 1993 atlas, vega_k93.fits, which can be found in the directory of HST standard spectra of the Calibration Data System (CDBS) at STScI. It covers a wavelength range of 0.009 – 160 µm with non-uniform sampling. In the near-IR, the spectrum is dominated by a strong continuum attenuated by several moderately strong absorption lines. To evaluate the potential impact of the absorption lines on the PDB, simulations were generated for an etalon tuning that places the absorption line due to Brackett ! (at ~1.95 µm) in the blue wing of the Airy function (Figs 8 – 10) on the short wavelength side of the non-functional region of the etalon. Similar simulations are shown in Figs 11 – 13 for a tuning that puts the same absorption line on the long wavelength side of the Airy function. As ! increases, the peak of the output profile (magenta in Figures 8 and 11) shifts to bluer wavelengths by ≲ 0.25% (Figures 9 and 12). The location of the absorption line in the Airy profile therefore does not affect the wavelength detuning. However, the corresponding change in the integrated counts is unusual. On the blue side of the nonfunctional region of the etalon, the integrated counts increase away from the optical axis by up to ~2%, while on the red side, they decrease by up to ~1% (Figs 10 and 13). How can we explain this behavior? For the small range of shifts of the profiles to the blue, all the output spectra should suffer a similar loss from the absorption lines. Hence, we hypothesize that to first order, the change in integrated counts is dominated by the interplay between the shape of the input spectrum and the edge of the blocking filter (neglecting the fine structure of the blocking filter and the narrowing of the Airy profile with !). • On the blue side of the non-functional region, as ! increases the Airy profile slides up the steep, blue SED of Vega and gains counts. But simultaneously, the profile loses counts in its far blue wing due to suppression by the blocking filter. The net result is a gain in counts as ! increases. The closer the Airy profile is to the blue edge of the blocking filter, the greater the loss in the blue wing, and the larger the overall increase (~0.2% vs ~2% in Figs 10 and 13). • On the red side of the non-functional region, the Airy profile is much closer to the blue edge of its blocking filter. In this case, the loss in the far blue wings dominates over the gain from the Vega spectrum, resulting in a net loss in counts. Again, the closer the profile is to the edge of the blocking filter, the greater the loss in the wings (~0.7% vs ~1% in Figs 10 and 13). However, other simulations with Vega show that a range of behavior can be obtained by moving the Airy profile across the blocking filter, especially near the edges. A rigorous analysis should include the competing contribution of the stellar continuum across the bandpass, the absorption lines, the fine structure in the throughput of the blocking filter, the proximity of the bandpass profile to the edge of the blocking filter, as a well as the PDB itself. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 10 - JWST-STScI-002190, SM-12 Figure 8 Simulations for ! = 0° on the short wavelength side of the TFI etalon’s non- functional region (top) and on the long wavelength side (bottom). The inputs are: Airy profiles (dashed black), Vega spectrum (green), and non-normalized blocking filter (cyan). The output spectrum (magenta = green x dashed x cyan) is shown in the bottom panels as well as the 2nd-order fit to the peak. The etalon gap is tuned to 3.9 µm so that the Br! absorption line falls in the blue wing of the Airy profile (uppermost panel). Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 11 - JWST-STScI-002190, SM-12 Figure 9 Flux-weighted wavelength shift for the simulations shown in Fig. 8 in absolute (top) and relative (bottom) wavelengths. Blue: shift of the profile on the short wavelength side of the non-functional region of the TFI. Red: same but for the long-wavelength side. Dashed: detuning described by Equation (1). Figure 10 Relative change in integrated counts due to PDB. Blue: output spectrum for the short-wavelength side of the TFI non-functional region; red: same but for the longwavelength side. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 12 - JWST-STScI-002190, SM-12 Figure 11 Simulations for ! = 0° on the blue side of the TFI etalon’s non-functional region (top) and on the red side (bottom). The inputs are: Airy profiles (dashed black), Vega spectrum (green), and non-normalized blocking filter (cyan). The output spectrum (magenta = green x dashed x cyan) is shown in the bottom panels as well as the 2nd-order fit to the peak. The etalon gap is tuned to 3.85 µm so that the Br! line falls in the red wing of the Airy profile (uppermost panel). Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 13 - JWST-STScI-002190, SM-12 Figure 12 Flux-weighted wavelength shift for the simulations shown in Fig. 11 in absolute (top) and relative (bottom) wavelengths. Blue: shift of the profile on the short wavelength side of the non-functional region of the TFI. Red: same but for the long-wavelength side. Dashed: detuning described by Equation (1). Figure 13 Relative change in integrated counts due to PDB. Blue: output spectrum for the short-wavelength side of the TFI non-functional region; red: same but for the longwavelength side. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 14 - JWST-STScI-002190, SM-12 3.2.3 A Planetary Nebula Spectrum Simulations were generated for the input spectrum of a planetary nebula. A synthetic spectrum of NGC 6543 (The Cat’s Eye Nebula) was created by modeling the H and He emission lines as Gaussian profiles with integrated fluxes and a weak, flat continuum tabulated in Hora et al. (1999) and Bernard-Salas et al. (2003). Other lines, such as Fe, are not included. For our purposes, an exact spectrum is not necessary: we simply want to characterize the consequences of the PDBs for an astronomical target with strong, narrow emission lines and a weak continuum. The results of the simulations are shown in Figures 14 –16. The etalon was tuned to a gap of 4.25 µm so that the Brα line at ~4.05 µm was situated in the red wing of the Airy profile on the long-wavelength side of the non-functional region of the etalon. Since many lines lie within the bandpass on the blue side, we measured the position-dependent change in wavelength by using the flux-weighted average (§3.1.2). We find that even though the detuning in wavelength across the FOV is small (Fig. 15), the corresponding variation in the counts can be very significant (up to a ~25% loss on the red side; Fig. 16) when the bandpass encompasses a strong emission line. This change is attributable to the steepness of the Airy function: as ! increases, the peak of the Airy function detunes to bluer wavelengths and in effect “slides” along the fixed emission line. As a result, large variations in the integrated counts under the narrow line are recorded. These changes are more pronounced near the peak of the bandpass, where the gradient d! d! is steepest. On the short-wavelength side of the etalon’s nonfunctional region, the effect is less pronounced since the loss in counts from the strong line at 2.17 µm in the red wing is partially compensated by the gain in counts from the line at 2.06 µm in the blue wing. This interpretation is confirmed when the etalon gap is tuned so that Brα falls in the blue wing of the Airy function on the long-wavelength side of the etalon’s non-functional region (Figs. 17 – 19). In this case, the change in counts shows a strong increase away from the optical axis, up to ~30% (Fig. 19), for the same small detuning in wavelength. Although this is opposite to the behavior seen when the emission line was placed in the red wing, it is consistent with expectations. In orbit, the user will likely place the target at the center of the field (! = 0°) and tune the etalon so the peak of the bandpass matches the wavelength of the emission line for maximum throughput. For Brα, the etalon gap would correspond to 4.276 µm. The simulation indicates that the change in the observed counts is significant, ~8% at " = 4.0°, since the Airy profile is shifted to the blue. Hence, if the source were observed again at another location in the field, the integration time would need to be increased accordingly to compensate for the loss in flux. ! Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 15 - JWST-STScI-002190, SM-12 Figure 14 Simulations for ! = 0° on the blue side of the TFI etalon’s non-functional region (top) and on the red side (bottom). The inputs are: Airy profiles (dashed black), planetary nebula spectrum (green), and non-normalized blocking filter (cyan). The net bandpass (blue = dashed x cyan) and the output spectrum (magenta = green x dashed x cyan) are shown in the bottom panels as well as the location of the flux-weighted average (vertical dotted line). The etalon gap is tuned to 4.25 µm so that Brα at ~4.05 µm falls in the red wing of the Airy profile. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 16 - JWST-STScI-002190, SM-12 Figure 15 Flux-weighted wavelength shift for the simulations shown in Fig. 14 in absolute (top) and relative (bottom) wavelengths. Blue: shift of the profile on the short wavelength side of the non-functional region of the TFI. Red: same but for the long-wavelength side. Dashed: detuning described by Equation (1). Figure 16 Relative change in integrated counts due to PDB. Blue: output spectrum for the short-wavelength side of the TFI non-functional region; red: same but for the longwavelength side. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 17 - JWST-STScI-002190, SM-12 Figure 17 Same as Fig. 14 but for a tuning of the etalon gap that places Brα in the blue wing of the Airy profile. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 18 - JWST-STScI-002190, SM-12 Figure 18 Same as Fig. 15. Figure 19 Same as Fig. 16. The change in integrated counts due to PDB shows a large increase as a function of !. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 19 - JWST-STScI-002190, SM-12 3.2.4 A High-Redshift Lyα Emitter For this application, the etalon is tuned to observe the Lyα emission line of a highredshift QSO. Such a program is under consideration by the TFI Science Team as part of their guaranteed observing time. We use the QSO spectrum of Telfer et al. (2002) as input. It is a combination of an HST radio-quiet composite below 1275 Å and the Sloan Digital Sky Survey median composite above 2000 Å and an average of the two in between. The H I opacity and Lyα forest are modeled according to the prescription of Madau (1995). At the high redshift of our simulations, the continuum blueward of Lyα is nearly completely attenuated; i.e., the Lyα forest imposes a “step” function on the SED. The simulations are shown in Figures 20–22. We assume the observer is searching for a QSO at a redshift of z = 13 so that the Lyα line will be located at the wavelength of highest resolution of the TFI, ~1.71 µm (Martel & Fullerton 2011). An etalon tuning of 3.35 µm places the peak of the “step” function just slightly blueward of the peak of the Airy function (Fig. 20). Once again, we find that even though the detuning in wavelength ! is small (≲ 0.04% at Lyα; Fig. 21), the corresponding change in counts is severe, up to ~25% at Lyα (Fig. 22). As for the simulations of the planetary nebula, we attribute this behavior to the sharpness of the Airy bandpass, particularly near its peak. Additional simulations offer more insights. If the etalon is tuned such that Lyα falls further in the blue wing of the Airy function (gap of 3.4 µm), then most of the weak continuum blueward of Lyα is outside the bandpass and the position-dependent changes are small, ≲ 1%. In fact, the counts increase with increasing ! because the Airy profile shifts to shorter wavelengths and includes increasing amounts of flux from the red wing of the broad Lyα profile. On the other hand, if the tuning is such that the Lyα peak is in the far red wing (gap of 3.3 µm), then a signfiicant decrease in counts is observed (up to 20%). However, unlike the previous simulations, the flux-weighted wavelength increases with increasing !. This is expected since the Airy function shifts further into the highly attenuated continuum of the Lyα forest as ! increases, which necessarily gives more weight to the red wing of the Airy function. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 20 - JWST-STScI-002190, SM-12 Figure 20 Simulations for ! = 0° on the blue side of the TFI etalon’s non-functional region (top) and on the red side (bottom). The inputs are: Airy profiles (dashed black), QSO spectrum at z = 13 (green), and non-normalized blocking filter (cyan). The net bandpass (blue = dashed x cyan) and the output spectrum (magenta = green x dashed x cyan) are shown in the bottom panels as well as the location of the flux-weighted average (vertical dotted line). The etalon gap is tuned to 3.35 µm so that Lyα falls in the peak of the Airy ! profile. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 21 - JWST-STScI-002190, SM-12 Figure 21 Flux-weighted wavelength shift for the simulations shown in Fig. 20 in absolute (top) and relative (bottom) wavelengths. Blue: shift of the profile on the short wavelength side of the non-functional region of the TFI. Red: same but for the long-wavelength side. Dashed: detuning described by Equation (1). Figure 22 Relative change in integrated counts due to PDB. Blue: output spectrum for the short-wavelength side of the TFI non-functional region; red: same but for the longwavelength side Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 22 - JWST-STScI-002190, SM-12 4. Implications of a Position-Dependent Bandpass These simulations indicate that the integrated counts of an astronomical source can vary significantly over the FOV of the TFI owing to the interplay between the PDB of the etalon and the SED of the source. These variations may have serious repercussions for both the implementation strategy and the results of an observing program. Here, we identify some issues that will require further consideration: 1. Dithers and mosaics: Except for coronagraphic observations, TFI observations will be dithered. For small dithers (a few pixels), no photometric correction is necessary within the pattern itself, although shifts along the concentric circle might be preferable to completely mitigate the effects of PDB. Of course, an overall correction to the flux may be necessary if the dithered source is located significantly away from the optical axis. For mosaics where several full fields of the TFI are observed with large offsets (say, ¼ of the FOV) but with some overlap, care must be taken to place the target(s) of interest at the same off-axis position in each field. Otherwise, the target will exhibit different integrated counts in each field, greatly complicating the analysis and the corrections. 2. Photometric corrections: For a given tuning of the etalon, any given target field will potentially contain hundreds of sources, some imaged in their continuum and others in an emission line. Generally speaking, the interaction between the PDB of the instrument and the SED of the source will affect each source differently. A significant problem for the observer is that the SED of each source is unknown a priori so any correction to be applied will involve some guesswork. One possible strategy is to observe the field at several wavelengths in order to construct the overall shape of the SED, although fine detail in the spectrum, such as narrow emission lines, will be difficult to capture. Some difficulties may also be encountered with coronagraphy in an emission line, such as a QSO, since the occulters are located near the edge of the FOV where the effects of are the most pronounced. Special photometric calibration may need to be implemented. 3. On-orbit commissioning and calibration: During the on-orbit commissioning and calibration campaigns for the TFI, the effects of the PDB can be characterized through observations of a source with a strong but narrow emission line, such as a planetary nebula or Be star. Changes in the position of the peak of the Airy function can be measured by scanning the etalon at multiple gaps (3 or 5, say) to locate the wavelength of the peak flux of the star at several off-axis radii. Changes in the shape of the bandpass can be tracked by using: 1) A star with a precisely known spectrum that is dominated by a continuum that is not very steep. The absorption lines should be weak and no emission lines should be present. A JWST spectrophotometric standard star would be a good candidate. By observing the star at several off-axis radii at a single tuning of the etalon, the effects of the changing shape of the bandpass can be mapped by measuring the total flux of the star at each of these field positions. The radial profile can then be compared with an optical model. Large deviations may signify a problem with the instrument’s optics. Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 23 - JWST-STScI-002190, SM-12 2) A planetary nebula spectrum might be preferable to a stellar spectrum for characterizing the position-dependent shape of the Airy function. If the planetary nebula is observed at several radii with a single tuning of the etalon that puts the narrow emission line in a wing of the bandpass, then extreme variations of up to ~10 – 25% will be measured in the integrated counts. However, measuring the flux of the extended, clumpy, or filamentary ionized regions of a nebula may prove more challenging and uncertain than the comparatively simple aperture photometry of a single Be star, for example. 5. Conclusions We have simulated the variations in wavelength and integrated counts over the FOV of the TFI for a variety of SEDs. We find that the effects of the etalon’s PDB can be significant and sometimes counter-intuitive. The largest effects occur when the incident SED has steep gradients, due to the presence of strong absorption features in a continuum (e.g., the Lyα forest in high-redshift Lyα emitters) or strong emission-lines (e.g., as seen against weak continua in planetary nebulae). Since the Airy function of the TFI’s etalon is also quite sharply peaked, the exact position of the spectral features associated with these gradients within the bandpass has a strong effect on the detected counts. The interplay between fine structure of the blocking filters and the shape of the Airy function can also be important. These results suggest that a judicious observing strategy and intensive modeling may be required to perform accurate photometry of sources in the FOV of the TFI. References Bernard-Salas, J., Pottasch, S.R., Wesselius, P.R., & Feibelman, W.A. 2003, A&A, 406, 165 Hora, J.L., Latter, W.B., & Deutsch, L.K. 1999, ApJS, 124, 195 Madau, P. 1995, ApJ, 441, 18 Martel, A.R. & Fullerton, A. 2011, An Introduction to the TFI Etalon, JWST-STScI002059, SM-12 Telfer, R.C., Zheng, W., Kriss, G.A., & Davidsen, A.F. 2002, ApJ, 565, 773 Check with the JWST SOCCER Database at: http://soccer.stsci.edu/DmsProdAgile/PLMServlet To verify that this is the current version. - 24 -