Predicted Countrates for the UV WFC3 Calibration Subsystem using Deuterium Lamps

Transcription

1 Predicted Countrates for the UV WFC3 Calibration Subsystem using Deuterium Lamps S.Baggett, J. Sullivan, and M. Quijada May 17,24 ABSTRACT Predicted WFC3 calibration subsystem countrates have been computed for a number of candidate deuterium (D2) lamps. Based on the calibration system component throughput estimates from Ball, the detector curves from the GSFC Detector Characterization Lab (DCL), and the lamp spectra measured at GSFC, use of any of the lamps in the calibration subsystem should provide sufficient flux to satisfy the CEI specification of 16.7 e - /s/pix in all filters blueward of and including F39N. This report summarizes the various lamps available and provides details of the procedure used for deriving the estimated countrates. Introduction Deuterium lamps manufactured by IST (Imaging and Sensing Technology) are planned for use in the WFC3 calibration subsystem (calsystem) to provide flatfields from ~2Å to 4Å. Initially, four D2 lamps were delivered to GSFC as possible candidates for the calsystem; since then, two have suffered weld problems and are no longer viable. As a result of the welding issues, IST revised their production procedures in order to fabricate more rugged lamps; a batch of five D2 candidate lamps with improved welds has been delivered to GSFC and one is currently undergoing vibration tests. This report summarizes the various lamps available for the calsystem, presents predicted calsystem countrates for filters Operated by the Association of Universities for Research in Astronomy, Inc., for the National Aeronautics and Space Administration

2 out to F39N, and provides details on the procedure used to derive the throughput estimates. Data Figure 1 presents spectra of the nine D2 lamps in question. The first four delivered were 3132, 3135, 3136, and 31298; the lamps with improved welds are 31774, 31775, 31776, 31886, and Of the first four lamps, number 3132 had one of the highest fluxes and was shipped to Ball for possible integration into the calsystem; the other high flux lamp (3135) suffered a broken anode weld during vibration testing and is no longer viable. Lamp was not considered a candidate for the calsystem, as not only did it have the lowest flux of the initial four lamps, it also exhibited some spectral features of unknown origin; since then, it too has suffered from a broken weld. The intermediate flux lamp 3136 was used in a lifetime test (Boucarut et al., 23; Baggett & Quijada, 23) and thus is also no longer a candidate for use in the calsystem. The five new design D2 lamps all have similar spectra but with relatively low output; their spectra cluster around the level of lamp 31298, the lowest flux lamp of the first four delivered. Figure 1: Spectra of nine D2 lamps. The top plot shows the entire wavelength range as measured; the bottom plot highlights the WFC3 calsystem wavelength regime. Radiance ( W/cm^2/nm) Spectral Scans All Viable D2 Lamps 3132 (Flight Candidate) 3135(Broken) 3136 (After Burn-in) 31298(Broken) Wavelength (nm).5 Radiance ( W/cm^2/nm) (Flight Candidate) 3135(Broken) 3136 (After Burn-in) 31298(Broken) Wavelength (nm) 2

3 Initial calsystem throughputs from Ball The Ball Systems Engineering Report OPT-6 (Sullivan, 21) provided predictions of the calibration system fluxes at the UVIS detector for filters out to F393N using an IST D2 spectral curve from the Cosmic Origins Spectrograph (COS) program. Table 1 in that report listed predicted countrates in each filter; these estimates were computed via an Excel spreadsheet, using NIST spectral radiance estimates, radiometric analysis by D.Ebbets, optical design as revised by M. Dittman, filter data from JPL, and additional updates by J.Sullivan in Jan 23. In that particular analysis, each calsystem observing mode was represented by a set of terms at the central wavelength of the filter in question (e.g., detector quantum efficiency, optics throughputs, and lamp intensity), which were multiplied by other necessary corrective factors (inclination of the detector to the line of sight, f/# of the light from the D2 lamp lens, and the total number of pixels) to derive the countrate estimates. Transfer of Ball calsystem throughput estimates into SYNPHOT For ease of use at STScI, the Excel spreadsheet computations were transferred into SYN- PHOT, a Synthetic Photometry package available as part of STSDAS/IRAF 1. SYNPHOT is commonly used at STScI for providing photometric calibration for HST data but it can also be used to provide countrate predictions. The SYNPHOT system consists of a set of throughput tables containing optical element throughput values as a function of wavelength, along with a graph table which defines which types of throughput tables to use for a given observing mode, and a component table, which provides specific names of throughput tables to use. To compute a predicted countrate, SYNPHOT identifies the necessary throughput tables for the specified mode, multiplies the files together, and sums the results across the bandpass. Once a SYNPHOT system has been established, it becomes relatively easy to swap out specific components and re-run countrate predictions. The current WFC3 SYNPHOT system, designed for estimating countrates from external sources, was used as a basis for generating a separate, calsystem-specific SYNPHOT system. Two terms are specified in the WFC3 SYNPHOT external observing mode: channel (IR or UVIS) and filter; these terms map to several tables for computing total throughput, namely, the HST OTA throughput table, WFC3 optics, filter throughput, and detector QE tables. For the calsystem, the HST OTA term is not needed; it was removed from the SYN- PHOT throughput path in the calsystem graph table. A new optics table was generated, to include the calsystem-specific components such as the UVIS beam splitter and the UVIS parabola, using the values given in the SER OPT-6-based Excel spreadsheet (Sullivan, 1. STSDAS is the Space Telescope Science Data Analysis System, which is part of IRAF, the Image Reduction and Analysis Facility, a general purpose software for the analysis of astronomical data 3

4 23). Those contributions, listed in the SER as a function of filter, were converted for SYNPHOT to a table of throughput as a function of filter central wavelength. Individual filter curves from the current SYNPHOT used for estimating external source countrates were kept as-is, since those contain the actual measured throughputs across each filter as a function of wavelength. For validation purposes, a SYNPHOT detector QE table and a D2 lamp spectrum were populated with the values used in the Excel spreadsheet (Sullivan, 23), also as a function of filter central wavelength. Once the calsystem SYNPHOT setup could be shown to reproduce the countrates given in OPT-6 and the Excel spreadsheet, the QE and lamp tables could be updated to reflect the most recent data available for the flight build. Since the Excel terms were given as a function of filter central wavelength while SYN- PHOT requires the tables to be populated out to the shortest and longest wavelengths in the system, some extrapolations in throughput and wavelength were necessary, from 2222Å down to 18Å and from 3933Å up to 45Å. A straight extrapolation to shorter wavelengths of the calsystem optics and the initial lamp curve gave rise to very steep curves, resulting in predicted SYNPHOT countrates significantly higher than the rates predicted in SER OPT-6, so the slope of those extrapolations were reduced somewhat; the extensions that were used in the calsystem SYNPHOT tables are shown in Figure 2 below. Figure 2: Initial SYNPHOT curves, based upon SER-6; dashed lines mark areas outside of which extensions were necessary (the QE required extrapolation on the blue side only). Calsystem optics, initial QE, and initial lamp curves are shown in the left, middle, and right plots, respectively. wfc3_uvis_d2 initial QE initial lamp 2.E E11.5 throughput 1.8E-1 throughput.4 flux 2.E11 1.6E E11 1.4E wavelength (angstroms) wavelength (angstroms) wavelength (angstroms) Using these extended curves along with the SYNPHOT filter throughput tables, a first set of SYNPHOT countrate predictions were computed. Some discrepancies were found between these initial SYNPHOT countrates and the Excel spreadsheet countrates, which were traced to differences between SYNPHOT filter throughput and/or bandwidth and filter throughput assumed in the Excel spreadsheet. Since the SYNPHOT filter tables are values measured at GSFC (Lupie and Boucarut, 23), the throughputs and bandwidths 4

5 from the SYNPHOT tables were used to update the Excel spreadsheet quantities and the Excel countrates were recomputed. Results Verification of Migration to SYNPHOT system One observing mode As an example, the countrate estimates for F28N are shown below. The prediction from the Excel spreadsheet (Sullivan, 23) is 6.7 e - /s/pix, computed using the following terms filter bandwidth = 42Å lamp intensity = 1e11 photons/s/å optics efficiency =.26 beam splitter =.42 mirror =.87 filter =.2 window =.92 QE =.47 (p/4) * ( 1 / f/# ) 2 = steradians cos(2) = (detector angle to the line of sight) 4K x 4K chip = pixels for a final countrate for the F28N D2 lamp calsystem observing mode of 42 1e = 6.7 In the initial calsystem SYNPHOT build, using the Excel spreadsheet preliminary lamp spectrum and QE, the resulting countrate was 62.1 e - /s/pix: calcphot wfc3,uviscalsys,d2,f28n spectrum=uvis.d2lamp.tab form=counts wave=wave.tab Mode = band(wfc3,uviscalsys,d2,f28n) Pivot Equiv Gaussian Wavelength FWHM band(wfc3,uviscalsys,d2,f28n) Spectrum: uvis.d2lamp.tab VZERO (COUNTS s^-1 hstarea^-1) or 62.1 e - /s/pix, within ~2% of the Excel spreadsheet value. The SYNPHOT path for F28N included the new wfc3 uvis calsystem lightpath table based on the contributions listed in the Excel spreadsheet, not including the lamp, the F28N filter throughput as a function of wavelength (from the pre-existing SYNPHOT system), the Excel-based detector QE table, and a spectrum (uvis.d2lamp.tab) based upon the lamp fluxes in the Excel spreadsheet. 5

6 All filters Calsystem countrate comparisons were done for all applicable filters by ratioing the countrate predictions from the Excel spreadsheet, using the updated bandwidths and filter throughputs, to the initial SYNPHOT build; the results, as a function of filter central wavelength, are presented in Figure 3. The left plot shows the first ratio results, using the SYNPHOT tables as defined to this point. Ideally, however, the SYNPHOT predictions should match the Excel predictions as closely as possible; to minimize the differences between the two systems, the initial ratios were fit as a function of central wavelength with a second-order polynomial and the slight curvature removed from the countrate ratios by including a correction based upon the fit in the SYNPHOT uvis calsystem table. Ratios of the resulting new SYNPHOT countrates to the Excel countrates are shown in the right plot, along with a fit to the new set of ratios. The inital ratios ranged from.89 to 1.5, with a standard deviation of ~.18; the final ratios ranged from.8 to 1.3, with a standard deviation of.15. Figure 3: Ratio of SYNPHOT to Excel spreadsheet countrate estimates for the D2 calsystem as a function of central wavelength. Left plot shows ratios using countrates from the first SYNPHOT build, while the right plot shows ratios using the adjusted SYNPHOT calsystem uvis table. 2 p ratio (SYNPHOT/Excel) 1 ratio (SYNPHOT/Excel) wavelength wavelength Updating CCD QE Curves and Lamp Spectra Once the SYNPHOT and Excel countrates were in reasonable agreement (see Figure 3), the new lamp curves and flight CCD QE curves were swapped into SYNPHOT. The spectra for the new lamps (shown in Figure 1), as measured in the optics lab at GSFC, were first converted from µw/cm 2 /nm to photons/sec/å, then multiplied by a factor of 5 to account for the GSFC detector intercepting only 2% of the lamp light. For QE response, the detector QE curves measured at the GSFC Detector Characterization Lab (Landsman, 22) were used. The DCL curves contain a correction for the quantum yield, to account 6

7 for the tendency for single UV photons shortward of ~34Å to generate more than one electron. In order to estimate calsystem countrates expected at the detector, the quantum yield correction was removed from the QE curves, following the prescription provided by DCL (Hill, 23); Figure 4 presents the adjusted QE curves used in SYNPHOT. The quantum yield effect is ~1.7 at 2Å, down to at (Hill, 23). To provide worst-case calsystem countrates, the lowest UV QE chips in each build were used: #178 for build 1 and #5 for the flight backup build 2. Figure 4: Flight detector QE curves; dashed lines are the QE curves from DCL which include a correction for quantum yield while the solid line has the correction removed and was used in the SYNPHOT estimates. 1 QE CCD178 1 QE CCD throughput.5 throughput wavelength wavelength 35 4 Countrate Predictions The resulting SYNPHOT countrate predictions for the five new lamps are shown in Figure 5 and tabulated in Appendix A. For completeness, the countrate predictions for are included as well, but note that due to a variety of factors, including a broken weld, it is no longer a flight candidate. Of all the lamps, both old and new, only the nonflight candidate failed the CEI specification of 16.7 e - /s/pix and then, only in two narrow band filters. The low countrates for this particular lamp may have been due to a conservative extrapolation of the spectrum out to 4Å; the scans of the new, more rugged lamps have data out to that wavelength and no extrapolations of those spectra were necessary. For flight build 1, which includes CCD178, all lamps exceed the CEI requirement by at least a factor of two. For the backup flight build, which includes CCD5, all of the new lamps easily meet the CEI specification and all but 1 lamp meet twice the CEI specification. Lamp 31774, in the F28N filter, falls just slightly below twice the requirement. 7

8 Figure 5: SYNPHOT countrate predictions for WFC3 UVIS calibration system as a function of filter number. Symbols cross, circle, plus, diamond, and asterisk represent the five new rugged lamps 31774, 31775, 31776, 31886, and 31887; box symbols represent the low flux lamp from the first batch of lamps (not a flight candidate). Solid line is the CEI specification; dashed line is twice the level of the CEI specification. 1 4 D2 lamp countrates, CCD D2 lamp countrates, CCD e-/sec/pix 1 2 e-/sec/pix filter filter Conclusions The D2 calibration system component throughput estimates from Ball have been migrated over to SYNPHOT, along with the detector QE curves from the GSFC DCL and the lamp spectra from the GSFC optics lab. Calsystem countrate predictions using the new SYN- PHOT tables show that all of the five rugged lamps meet the CEI specification of 16.7 e - /s/ pix in filters blueward and including F395N, in either flight build1 or build2. Furthermore, for build1, all lamps exceed at least twice the CEI specification; for build2, all but one lamp (31774) exceed the CEI requirement by a factor of two in all applicable filters. Acknowledgements We thank D.Ebbets for helpful discussions of D2 lamps and the WFC3 calsystem. 8

9 References Baggett, S., and Quijada, M., Lifetime Test of Deuterium Lamp for the WFC3 Calibration Subsystem, ISR 23-11, Nov 23. Boucarut, R., Garcia, K., and Kutina, R., Lifetime Test Plan for the HST Wide Field Camera 3 Calibration Deuterium Lamp, April 23, GSFC internal document. Hill, R., priv.comm, Nov 23. Hubble Space Telescope Wide Field 3 Contract End Item Specification, May 2, HST Library #TM-27755, section , pg Landsman, W., tables of DCL UVIS detector QE curves, priv.comm, June 22. Lupie, O. and Boucarut, R., WFC3 UVIS Filters: Measured Throughput and Comparison to Specifications, WFC3 ISR 23-2, Feb 23. Sullivan, J., UVIS Calibration Subsystem Predicted Flux at UVIS Detector,, Ball SER OPT-6, July 21. 9

10 Appendix A. Table 1. Estimated countrates, in e - /s/pix, for D2 calsystem. Listed are the countrates for the lowest and highest flux flight candidate lamps and 31887; for comparison, countrates are listed as well though it is no longer flight candidate. For signal to noise calculations, countrates can be corrected for the quantum yield effect by dividing by the factor in the last column. filter lamp lamp lamp quantum yield ccd 178 ccd 5 ccd 178 ccd 5 ccd 178 ccd 5 factor f218w f225w f232n f243n f275w f28n f3x f336w f343n f373n f378n f387n f39m f39w f395n