In-flight Characterization and Calibration of the Juno-ultraviolet Spectrograph (Juno-UVS)

Juno-UVS is composed of two separate assemblies. The instrument sensor housing, which holds the telescope, detector, and detector electronics is located on the spacecraft's deck, and is looking radially outward from Juno's spin axis (Bolton et al. 2017b). The command and data handling electronics (C&DH) and power supplies are located in an electronics housing within the spacecraft vault to protect them from radiation. The two parts are linked by approximately 4 m of cable.

An overview of the telescope assembly and detector is provided in Figure 1 (left) and shows the trajectory followed by photons in color, with a different color after each reflection on an instrument optical surface. The photons entering the telescope first hit the scan mirror, which redirect them through a 4 × 4 cm² aperture toward an off-axis parabolic mirror (OAP). The OAP focuses the photons on a dog-bone-shaped slit, represented in Figure 1 (right). The dimensions of the slit are 255 × 02 for each of the wide outer sections, and 2° × 0025 for the narrow slit of the central section. The photons then hit the toroidal diffraction grating, mounted in a Rowland circle configuration with the slit and the microchannel plate (MCP) detector. This light is then diffracted and focused onto the detector. The aluminum scan mirror, OAP, and diffraction grating are coated with an aluminum and magnesium fluoride overcoat (Al/MgF₂) to optimize reflectivity in the UVS spectral bandpass (e.g., Gladstone et al. 2017a, for a more detailed discussion).

Zoom In Zoom Out Reset image size

The detector consists of three MCPs in a Z-stack configuration that are read out by a cross-delay line (XDL) anode. The Z-stack is cylindrically curved in order to match the 150 mm Rowland circle to optimize the spectral focus. The input surface of the Z-stack is coated with an opaque, solar-blind, UV-sensitive, Cesium Iodide (CsI) photocathode, providing good quantum detection efficiency over the UVS spectral bandpass (Siegmund 1999). The choice of that instrument design was made in order to achieve UVS science goals, i.e., observing and characterizing Jupiters FUV auroral emissions, while taking advantage of the heritage from previous successful instruments. The detector-pixel readout format is 2048 × 256 (spectral x spatial) over the 4 × 3 cm² surface, and with an active detection array of 3.5 ×1.8 cm². Each pixel is 25 μm and 89 μm in the X- and Y- directions, respectively. This corresponds to 0.1020 nm ±0.0004 nm in the spectral direction (X) and 0.041° ± 0.001° in the spatial (Y) direction. The detector electronics amplify the signal coming from the MCP, and convert the detected output pulses to locate the X (spectral) and Y (spatial) coordinates of the detected events. Only those analog pulses output from the MCP that have amplitudes above a set discriminator level are processed and converted to pixel address locations (Stern et al. 2008; Gladstone et al. 2017a). In order to monitor the health of the instrument, i.e., the output gain of the detector at a given high-voltage (HV) level, the pulse heights are recorded for each event. The pulse height distribution (PHD) is produced by histogramming the pulse heights. The shape of a PHD can help to identify the sources of detected counts, UV photons, electrons, ions, or gamma-ray photons (Stern et al. 2007).

The C&DH controls the high-voltage power supply (HVPS) level (and by default the detector gain), the detector heater, and actuator activation mechanisms, as well as the scan mirror mechanisms. Among other things, it also monitors the detector count rate, voltage, and strip current so that it can protect the instrument against anomalous conditions.

The spatial point-spread function (PSF) was measured during the high-voltage commissioning (HVC) and the periodic maintenance 1 (PM1). The description of the operations performed during these sequences were described in an earlier paper (Greathouse et al. 2013). The PSF is expected to vary along the slit because the spatial focus varies depending on the slit position. Previous work from Greathouse et al. (2013) showed that the PSF along the slit, i.e., parallel to the slit direction, varies almost linearly from ∼0.15° up to ∼0.29°. The PSF across the slit, i.e., in the spin direction, is best in the narrow part of the slit, where it is ∼0.19°, and is ∼0.25° over the wide slits. The spectral resolution of UVS is also slit-dependent. It is best in the narrow slit (∼1.3 nm), and worst at the outermost ends of the wide slits (∼3 nm). The main characteristics and performances of the Juno-UVS instrument are summarized in Table 1.

Table 1.  Overview on the Instrument Characteristics and Performances

Spectral Range	68–210 nm
Point source FWHM along the slit	∼0.15°–0.29° along the slit
Point source FWHM across the slit	0.25° (wide slits) 0.19° (narrow slit)
(i.e., in the spin direction)
Spectral resolution	∼1.9–3.0 nm (wide slits)
	1.3 nm (narrow slit)
Field of view	0.2° × 2.55° + 0.025° × 2° + 0.2° × 2.55° $\to$ dog-bone-shaped
Field of regards	360° × 60°
	(2 rpm and scan mirror allows ±30° from spin plane $\to$ half the sky is accessible)
Spectral (X) plate scale	0.1020 nm ± 0.0004 nm
Spatial (Y) plate scale	0.041° ± 0.001°
Effective area (this work)	0.30 ± 0.03 cm2 @ 125 nm, 0.15 ± 0.02 cm2 @ 140 nm, and 0.05 ± 0.01 cm2 @ 160 nm

Note. Information gathered from Gladstone et al. (2017a), Greathouse et al. (2013), and this work.

Download table as:  ASCIITypeset image

The versatility available in Juno-UVS commanding makes it usable for a wide range of observations. During much of each orbit Juno-UVS is off. When on, the different observations performed are, typically

• 1.  
  perijove observation,
• 2.  
  synoptic observation,
• 3.  
  calibration observation,
• 4.  
  and radiation monitoring.

Perijove observation. The perijove observations are nominally performed for about 10 hr centered on perijove. The spatial resolution on Jupiter changes drastically during this period, and extensive use of the scan mirror is used to target different auroral features on Jupiter. The strategy in these is to optimize the data acquired, while avoiding high background radiation regions. Radiation models are used in order to avoid as much as possible taking data in the high radiation regions (Becker et al. 2017). The interested reader can refer to the work of Kammer et al. (2018) for further information on the strategies adopted to mitigate the impact of the radiation on the science return.

Synoptic observation. The synoptic observations are performed when Juno is located in the outer part of the orbit. During these observations, Jupiter appears as ∼1° to UVS meaning the instrument can still resolve the disk, but only with a few resolution elements making these observations useful for auroral power calculations rather than auroral imaging studies (e.g., Gladstone et al. 2017b). Since the plane of Juno's orbit is inertial, the orientation of that plane evolves from a dawn/dusk orbit (PJ1) to a midnight/noon orbit (PJ22) due to the orbit of Jupiter about the Sun. This evolution means that UVS does not have the capability to observe Jupiter in the synoptic configuration from PJ10 (2017 December 16) up to ∼PJ33 (2021 April 15). Since PJ10, regular half-sky scans have been performed instead of the synoptic observations, where the full range of the scan mirror is used in order to map half of the sky.

Calibration observation. Calibration observations are performed regularly to evaluate the health of the instrument. Four scan mirror pointings are chosen for the calibration observations so that ∼2.5 hr of time may be spent at each pointing, depending on the instrument's parameters (HVPS). For each swath of the spacecraft, a given point source is observed during typical ∼17 ms and ∼2 ms integration times through the wide and narrow slits, respectively. This gives an effective integration time on any one star of ∼4 s for the wide slit and 0.6 s for the narrow slit, when 2.5 hr of observations are co-added. See the work of Greathouse et al. (2013) for a more in-depth discussion.

Radiation monitoring. Finally, the radiation observations are performed to monitor the radiation environment by running the instrument with the aperture door closed and by recording the events triggered by high-energy electrons, gamma-rays, and galactic cosmic rays passing through the instrument shielding. Figure 2 represents a typical orbital sequence performed during PJ8 (2017 September 1).

Zoom In Zoom Out Reset image size

An inspection of the total number of events recorded by UVS gives a measure of how many UV photons and radiation events the instrument has experienced since its launch. Figure 3 summarizes the fluence experienced by the instrument since its launch. This quantity was calculated as

$$\begin{eqnarray}&&{\rm{\Phi }}=\displaystyle \frac{\epsilon \times e\times \sum \mathrm{ACs}}{{\sigma }_{\mathrm{MCP}}},\end{eqnarray} \tag{ 1 }$$

where Φ is the charge depletion, i.e., the amount of charge extracted from the glass plates due to photon counting, in C cm⁻², is the overall detector gain ( = 2 × 10⁷, from Gladstone et al. 2017a), and σ_MCP is the area of the MCP (3.5 × 1.8 cm²). The last term of the right-hand side of the equation represents the summation over the number of events recorded in the analog list, i.e., the analog counts (ACs).

Zoom In Zoom Out Reset image size

The charge depletion is presented here per data set to provide an inter-data set comparison. It does not represent the total charge depletion experienced by the detector, which can instead be computed by the cumulated charge depletion since launch. During cruise, from launch up to mid-2016, UVS performed regular routine observations, such as periodic maintenances, HVC, or dedicated observations (e.g., Earth flyby, observations of the ISON comet). The different types of observations are color coded and dark green denotes the JC64 approach sequence that ran for 26 days straight. The time period corresponding to each of the event discussed are presented in Table 2.

Overall, the amount of charge depleted per data set stays fairly constant during the cruise to Jupiter, around 5 × 10⁻⁵ C cm⁻², with quite a large scatter. These variations are related to differences in the pointings used and the decrease of the Lyα interplanetary medium (IPM) background as Juno travels farther from the Sun (Broadfoot et al. 1989).

The JC64 sequence was a special sequence of almost continuous observation of Jupiter during approach, taken from 2016 June 3 to June 29 (Gladstone et al. 2017b). It used a small set of scan mirror pointings to keep Jupiter well centered in the UVS lower wide slit. The variation in the charge depletion over the JC64 observation period cannot be attributed to variation of the Lyα IPM background since the spacecraft/Jupiter/Sun geometry was nearly constant over this period. However, the overall count rate was measured to have decreased by a factor of 2 over this 26-day-long observation period, mostly due to scrubbing and loss of sensitivity at Lyα.

Once at Jupiter, the orbital operations clearly show extensive use of the detector. The charge depletion of the detector during the PJ observations is ∼40 times that of any other observation performed. The large scattering of the PJ observation points are mostly due to the orientation of Jupiter's magnetic field, which controls the amount of counts UVS receives due to high-energy electrons.

Numerous dark observations have been performed throughout the mission. They consist of running the instrument with the aperture door closed, and provide an assessment of the detector performance. Such an evaluation of the detector performance through selected dark observations is presented in Figure 4. The background level experienced by the spacecraft may vary during each dark observation. Each dark observation in Figure 4 has been normalized by the averaged count rate within the yellow box shown on the left-hand side of the detector.

Zoom In Zoom Out Reset image size

Previous work showed that the detector dark counts exhibited a long arcuate feature early in the mission, although it was thought not to cause problems given its low count rates compared to observational sources of interest (Greathouse et al. 2013). The feature disappeared during the Earth flyby observation data set, possibly drowned out by the high level of radiation experienced. It then reappeared during cruise and definitely disappeared once into the Jovian system. Another likely related feature briefly appeared when the potential difference of the HVPS was increased for the first time at −204 V below the nominal HVPS level, during a calibration observation taken around PJ5. The new feature was observed during one calibration data set, and never reappeared later (as of PJ14), even when testing the instrument response at a higher HVPS level.

The depletion in detector counts seen at Lyα is characterized by the vertical hourglass that stands out of Figure 4. Located near the center left of the detector spectral range, it progressively shows a dip in the number of counts recorded at these wavelengths, and clearly reveals the shape of the slit during the Jupiter operation.

The extensive use of UVS requires close monitoring of the health of the instrument. Numerous stars have been used throughout the mission as calibration standards to inspect the sensitivity of the instrument as a function of time.

For example, Figure 5 presents several UVS calibration spectra of the UV-bright star A Pup (HD 54893). This star was observed multiple times over more than 3 yr through the wide part of the slit, at the nominal HV level (−4177.5 V). Only the 2014 measurements shows sensitivity to Lyα. The right-hand side of the plot shows the PHD of the photons recorded for every visit, with the Lyα region and longer wavelengths separated. A minor design flaw in the detector electronics causes pulse heights that would otherwise fall in bins 17–32 to instead all stack up in bin 17, without affecting the counting efficiency. This flaw is the reason some PHDs show a large spike in bin 17. In order to be able to detect a photon, the pulse height triggered by a photon must lie within the range of the discriminator. After an extensive use of the instrument, the glass plates are scrubbed of charge in the areas of highest integrated fluence, making future events measured in these areas have a lower pulse height, also called gain sag. This phenomenon can be tracked by monitoring the PHD over time. At the nominal HV level, UVS lost its sensitivity to Lyα through the wide slits prior to reaching Jupiter. UVS is however still sensitive, as of PJ14, at the longer wavelength photons and through its wide and narrow slits. As the mission progressed, the PHD of longer wavelengths were observed to gradually shift to lower bins. In order to avoid losing sensitivity at the longer wavelength, the HV of the instrument was regularly raised over the mission. The reader interested in the effect of gain sag can refer to the work done on, e.g., the Hubble Space Telescope Cosmic Origins Spectrograph (HST-COS; Osterman et al. 2011; Sahnow et al. 2011).

Zoom In Zoom Out Reset image size

Note that the UVS settings for the curve B of Figure 5 were such that the instrument was masking the Lyα region. Note also that, although HD 54893 was included in the General Catalog of Variable Stars (GCVS; Samus' et al. 2017), it was ranked as a suspected variable (Kholopov 1982). A closer look to the published observations shows little to no variations (Jerzykiewicz & Sterken 1977).

Due to the extensive sky coverage, long periods of time away from PJ, and large available downlink, Juno-UVS is able to perform extensive tracking of instrument sensitivity. This is accomplished by looking at a handful of stars throughout the mission, over a wide range of HV values and well-sampled temporal cadence. To illustrate, Figure 6 shows how the location of the PHD peak evolves as the mission progresses. Since the sensitivity of UVS is expected to decrease with the instrument usage, this was plotted as a function of cumulative charge depletion since launch, making the x-axis nonlinear in time. The star observed was HD 28873 (δ Cae), a 5.07 V-magnitude B2IV-V star, for which only the wide slit measurements are shown here. The cumulative charge depletion at the time of each PJ is also represented through the vertical dashed line. Since the fluence experienced by UVS varies at each PJ (see Figure 3), the PJ positions are not equally distributed in the cumulative charge depletion space.

Zoom In Zoom Out Reset image size

HD 28873 was observed over a wide range of HV levels, displayed by the labels on the left-hand side of the graphic, and showing the difference in the HV level with respect to the nominal one. The numbers reported in parenthesis next to the HV level denote the number of measurements performed at that HV setting. Most of the observations have been performed at the nominal HV setting. Shortly after PJ3, several calibration observations at a lower voltages were performed in order to quantify the instrument sensitivity at these HV settings. Consequently, the PHD was shifted toward the lower discriminator level. Calibrations at several higher voltage settings above the nominal one were also performed throughout the mission to pre-test the instrument response at future settings. The FWHM of a healthy PHD is about 5 bins. In order to optimize the detection efficiency, the UVS parameters should remain such that the PHD maximum does not go lower than bin number 5–8. Consequently, the potential difference of the HVPS of UVS was increased by −51V, −102V, and −153V below the nominal HV level prior to PJ8, PJ10, and PJ13, respectively. Proceeding to these changes hence increases the absolute value of the voltage applied, keeping the instrument detection efficiency steady over time.

The PHD peak location of the upper wide slit data points are systematically shifted down by ∼1–2 bins, for a measurement over the same star at a given time and voltage setting. Consequently, data points corresponding to lower wide slit measurements are shown by plus signs, while upper wide slit measurements are shown by triangles. The colored straight lines represent fits of the PHD peak location as a function of charge depletion, at a given HV setting. The fits on the lower and upper wide slit measurements, when enough data was available, are shown with solid and dashed lines.

This operation was performed for a number of stars that fall into a reasonable range of magnitude (V-mag between 4.5 and 6.5), and that have been detected multiple times at nominal voltage. The position of the PHD peak was tracked for all of these stars and fitted as a function of charge depletion, for a given voltage setting. At the nominal voltage level, 46 stars fulfilled these criteria: 31 stars were measured through the lower wide slit and 28 stars through the upper wide slit. Figure 7 shows how the PHD peak location decreases over time for this subset of stars. Measurements made through the lower wide slit are shown with solid lines, while upper wide slit measurements are shown with dashed lines. Several outliers have been removed by excluding the extremum of the fits (i.e., minimum and maximum values for the slopes of the lines). The overall charge depletion between the two wide slits seems consistent with one another, given the scatter on individual star charge depletion measurements. Note that the upper wide slit should be used first to track down the moment where the detector will start missing photons, since it exhibits an offset in term of PHD peak location of about 1.7 bins.

Zoom In Zoom Out Reset image size

The rate of charge depletion (i.e., the slope in Figure 7) can be compared for the measurements made at different voltage settings. Figure 8 represents the PHD depletion rate for the three voltage settings used during the PJ observation, as of PJ14. The depletion rate experienced by UVS at different HV levels stays constant for the different settings, within uncertainties.

Zoom In Zoom Out Reset image size

Observations made by two UV observatories have been used as a reference: the International Ultraviolet Explorer (IUE) and the Hubble Space Telescope (HST). The HST observations used are part of the Space Telescope Imaging Spectrograph (STIS) Next Generation Spectral Library⁵ (Heap & Lindler 2010; Lindler & Heap 2010), as well as observations compiled in the CALSPEC⁶ library.

One of the most extensive stellar UV-spectral databases comes from the IUE mission. It recorded over 100,000 UV spectra between 1978 January 26 and 1996 September 30. The satellite observed a variety of astrophysical bodies, ranging from comets, planets, stars, and supernovae to galaxies, in the 115–335 nm range, at high (0.01–0.03 nm) and low (0.6–0.7 nm) spectral resolution.

The IUE spectral database was the prime source used to calibrate Juno-UVS. All of the spectra recorded by IUE for O, B, and A stars were downloaded from the MAST database,⁷ leading to a total of about 45,000 different spectra. A closer inspection at specific stellar spectrum shows a great variability between measurements recorded of the same star. The IUE database, which includes a large number of observations taken at different times, with different observing modes and instrument parameters, was refined in order to only keep the spectra that were thought to be reliable and usable for calibration.

The IUE had two spectrographs, the short-wavelength spectrograph, operating between 115 and 200 nm, and the long-wavelength spectrograph, operating between 185–335 nm. Two cameras per spectrograph were developed, the prime and redundant cameras for the short-wavelength spectrograph (SWP and SWR, respectively), and similarly for the long-wavelength spectrograph (LWP and LWR). The SWR was not fully functional during the mission, and no spectra recorded with it were used within our process. IUE was capable of observing sources through two different apertures: the large aperture (∼10'' × 20'' ellipse) and the small aperture (∼3'' × 3'' circle). The size of the image for a point source in the aperture plane was 4'' × 4'', making the absolute flux calibration of the small-aperture observations difficult (Garhart et al. 1997; Massa & Fitzpatrick 2000). Consequently, only the large-aperture observations were used. IUE data recorded under abnormal conditions were not used. Several flags in the FITS headers were written to report these anomalies. Therefore, the IUE fits files recorded with at least one of these header flags written as affirmative value (i.e., YES) were discarded: ABNBADSC, ABNHISTR, ABNHTRWU, ABNMICRO, ABNMINFR, ABNNOSTD, ABNOTHER, ABNREAD, and ABNUVC. The spectra of variable stars that were found in the GCVS (Samus et al. 2017) were also discarded. The International Ultraviolet Explorer Data Analysis Center (IUEDAC) IDL Software Libraries⁸ were used to read in the IUE fits files, which involve the New Spectral Image Processing System (NEWSIPS) calibration pipeline (Nichols & Linsky 1996). Further flux calibration correction was applied on the low-dispersion data, following the investigating from Massa & Fitzpatrick (2000). Only the observations recorded using a single exposure mode were used. The observations recorded with the trailed exposure mode were discarded, as well as the ones designed with an offset in the X–Y axis of IUE's Fine Error Sensor (FES; see Garhart et al. 1997).

A recent correction for the IUE low-dispersion flux calibration was recently published by Bohlin & Bianchi (2018) to better match the CALSPEC library. We closely compared the refined IUE database using that correction with the CALSPEC and NGSL database. Although it seems clear that this correction provides a better match of the IUE observations with the CALSPEC library, the agreement between the IUE and NGSL however degraded at the same time. In this first version of the Juno-UVS calibration, the correction published by Bohlin & Bianchi (2018) is not accounted for.

When applying these selection criterion, the remaining IUE database used for calibration totals about 1800 different stars. Some of them have been observed multiple times, and, by cross-comparing the spectra for the given stars, several measurements still exhibited discrepancies and/or significant variabilities. For instance, the measurements showed systematic discrepancies in the outermost part of the spectrograph bandpass, particularly the LWP/LWR spectrographs. We limited the bandpass of the SWP spectrograph to be 115–200 nm (Nichols & Linsky 1996), and the LWP/LWR spectrographs to be 185–335 nm (Massa & Fitzpatrick 2000).

In order to build a catalog of stellar measurements, we degraded this refined IUE database to the Juno-UVS resolution. Further refinements have been applied (i) when three or more measurements of the same star have been recorded by excluding the IUE measurements that appeared statistically different from others, (ii) by discarding measurements recorded with the shortest exposure time, (iii) by cross-comparing the IUE spectrum in the overlapping spectral region of the SWP and LWP/LWR and discarding the measurements that appeared significantly off, (iv) by discarding single measurements presenting a large gap in wavelength, and (v) by cross-comparing with HST observations.

When three or more IUE measurements taken with the same spectrograph (SWP or LWP/LWR) were recorded, the averaged flux and standard deviation in several specific wavelength regions were computed. Measurements further apart than one standard deviation have been discarded. When five or more measurements with the same instrument were recorded, the bottom 15% in terms of exposure time duration have been discarded. When both short- and long-wavelength measurements for the same star exist, the averaged flux in the overlapping spectral region of 190–195 nm were used to discard the long-wavelength measurements that were ∼10% or more different from their short-wavelength counterpart. Several measurements presented a large gap in terms of wavelength coverage and they were similarly discarded. Finally, HST observations were used to cross-compare the IUE observations in the 170–185 nm region. When the relative difference with the IUE observations exceeded 80%, only the HST observations have been retained for the calibration. An example of the IUE observation selection process is illustrated in Figure 9.

Zoom In Zoom Out Reset image size

In this section, we describe how the effective area of UVS has been retrieved using the IUE/HST spectra database established previously. As shown in Figure 3, most of the instrument sensitivity at Lyα in the wide slits was lost prior to JOI. We will first discuss how the non-Lyα calibration was performed, and then we briefly discuss the Lyα calibration.

4.2.1. Non-Lyα Calibration

Because Juno is a spinning spacecraft that nominally spins at 2 rpm, it is not possible to sit and stare at specific stars known to be reliable for calibration purposes. Instead, the mirror position is defined such than Juno-UVS will observe a handful of stars over one spin. Several spins of data recorded at the same mirror position were co-added to increase the signal-to-noise ratio. During each spin, a point source will be observed for about 17 ms in the wide part of the slit and 1.9 ms in the narrow part of the slit, when the mirror position is nominal, i.e., when the UVS boresight direction is within the spin plane. The integration timescales up as 1/cos²α_i, where α_i is the angle between pixel i along the slit and the Juno spin plane—the square of the cosine is due to both the circle described on the sky being smaller and the star crossing the slit at an angle.

The following calibration algorithm has been written into an automated IDL calibration code and is illustrated in Figure 10. One single continuous chunk of the Juno-UVS observation is first considered, implying that it was recorded with the same instrument parameters without any temporal gaps and at a single mirror position. The results are mapped and displayed in Figure 10(A). The routine starts with the brightest star in the map and moves toward dimmer stars as it proceeds. Several tests are performed on the selected stars to assess whether they are good for calibration (see Figure 10(B)). If it fails any of the tests, the star is masked and the algorithm proceeds to the next one. These tests range from checking that the stars do not have an abnormal size by fitting a Gaussian, was not recorded too near the ends of the slit, is not too close or merged to other bright stars, and finally that it can be attributed to a known star within a certain angular distance from the detected centroid. Once the star passes these tests, the background subtraction is done by finding a nearby region, devoid of light contamination from neighboring stars, preferentially recorded in a detector region that is as close as possible from the one of the considered star (see Figure 10(C)). Results of the extracted stellar and background fluxes are shown on Figure 10(D), in black and red, respectively, and correspond to the black and red boxes of Figure 10(C). Finally, if a measurement exists for that star in our refined IUE/HST observations, the effective area of Juno-UVS is calculated (Figure 10(E)).

Zoom In Zoom Out Reset image size

That algorithm is automatic and generates typical plots such as the one in Figure 10 for all detected stars. It has been run on all the synoptic and calibration observations (recall Figure 2). Up to PJ14 (2018 July 16), around 8700 individual stellar spectra have been extracted, 7500 and 1200 through the wide and narrow slits, respectively. Many stars have been observed multiple times. Only spectra for which previous IUE/HST observations exist have been retained, meaning that the effective area of the instrument has been retrieved that same number of times.

Bright stars cause localized gain suppression on the detector when observed through the wide slits, which makes the detector miss a fraction of the incoming photons. Only a subset of stars can be used for calibration. Furthermore, since the throughput of the wide slit versus the narrow slit is different, the range of the V-magnitude of the stars retained for calibration depends on the slit considered. Previous work of Greathouse et al. (2013) used only a handful of stars to perform an early calibration during the HVC. They used stars with a V-magnitude around 3.15–4.7 and around 2–2.5 for the wide and narrow slits, respectively. In this work, we used stars in the V-magnitude range of 4–6 and 1–3 for the wide and narrow slits, respectively.

Every retained effective area measurement was subsequently fitted following the procedure described on Figure 11. We fitted the 125–220 nm region with a third order polynomial function. The 115–119 nm region was found challenging to fit due to large uncertainties on the measurements. In addition, there is a lack of reliable calibration data at wavelengths shorter than 119 nm over the large set of stars observed by UVS. We fitted the wavelength shorter than 119 nm using a theoretical effective area expected and rescaled to fit the mean measured effective area in the 115–119 nm range. The theoretical effective area was calculated considering the optical design of Juno-UVS and assuming an MgF₂ coating on all of the optics (Gladstone et al. 2017a). Using a subset of the brightest stars detected by UVS, we also compared our effective area calculation at these short wavelengths where the extrapolation is done using Kurucz models (Kurucz 1979). Our results are in very good agreement considering the measurement uncertainties, and the artificial nature of the calibration. The same exact procedure was performed on the measured errors.

Zoom In Zoom Out Reset image size

This procedure was performed for all of the effective area measurements that fulfilled the aforementioned criteria. Figure 12 provides a summary on the overall retrieved effective areas as a function of time and HV setting. Each column denotes a different part of the slit: lower wide slit (left), narrow slit (middle), and upper wide slit (right). Each row displays an ensemble of the retrieved effective areas at a given HV level and a given time. The number of measurements used for each panel is shown in parenthesis. While the white solid lines denote the mean effective areas, averaged from every single fitted effective area measurements (recall Figure 11), the white dashed and dotted lines were calculated using these 2D histogram distributions by finding the center and the standard deviation of the distribution for each wavelength bin.

Zoom In Zoom Out Reset image size

Figure 12 suggests than that are no obvious changes in the instrument sensitivity at non-Lyα wavelengths across the detector and over time. A closer inspection on a time series of the upper wide slit retrieved effective areas, i.e., the one that could potentially first shows signs of gain loss, is presented in Figure 13. Given the uncertainties on the effective area measurements, the upper wide slit does not exhibit any temporal variability in the retrieved effective area when compared to the averaged one.

Zoom In Zoom Out Reset image size

4.2.2. Lyα Calibration

A procedure similar to the one used for the non-Lyα region was carried out here, with a few exceptions. First, the star selection algorithm previously used was run on brightness maps integrated over the 119–123 nm region. Due to the loss of sensitivity at Lyα already recorded prior to JOI, only the calibration in the narrow part of the slit was considered here. A very limited number of stars were found to be valid for calibration at Lyα, making the process challenging to achieve, leading to greater uncertainties and a lower level of confidence than for the non-Lyα wavelengths.

Only stars having significant flux at Lyα have been retained to perform the calibration at these wavelengths. Table 3 summarizes the stars used for the Lyα calibration of the narrow slit of UVS. Several of these stars have previously been classified as variable. The way Juno-UVS takes data, i.e., one limited exposure at each 30 s spin, and because a large number of spins have to be co-added to build signal, the impact of using few variable stars in the Lyα calibration should be minimal.

Table 3.  Stars Used for the Lyα Calibration

Star Name	V-magnitude	Spectral Type	Number of Observations
κ Ori (Saiph)	2.07	B0.5Ia	65
CMa (Adhara)	1.50	B1.5II	43
α Gru (Alnair)	1.71	B6V	24
ζ CMa (Furud)	3.00	B2.5V	20
α Pav (Peacock)	1.92	B2IV	12
η UMA (Alkaid)	1.86	B3V B	9
σ Sgr (Nunki)	2.07	B2V B	2
γ Ori (Bellatrix)	1.64	B2V C	2
α Leo (Regulus)	1.40	B8IVn	2

Download table as:  ASCIITypeset image

The effective area at Lyα on the narrow slit shows some degradation over time. To investigate this effect, a time-dependent retrieval of the effective area was performed. For each perijove, that quantity was retrieved using all of the stars measured at the voltage setting used during that PJ, and over a period ranging from +/− 1 PJ (∼106 days) around the PJ of interest. Figure 14 shows how this effective area evolved as a function of time. During cruise the measured effective area at Lyα was 0.21 cm², and had degraded down to 0.07 cm² by PJ7.

Zoom In Zoom Out Reset image size