3.1.

Stellar EUV Irradiance

Optical and near-infrared photons heat the surface and troposphere. NUV (1800 to 3200 Å), FUV (912 to 1800 Å), and x-ray (5 to 100 Å) photons are absorbed in the middle and upper atmosphere where they photo-dissociate molecules and ionize heavy elements. EUV photons (100 to 911 Å) are absorbed high in the atmosphere (i.e., in the thermosphere) where they ionize atoms and molecules. Liberated electrons collisionally heat the surrounding gas, increasing the scale height of the atmosphere, driving ambipolar ion mass loss, and potentially leading to the formation of a hydrodynamic outflow. Rapid atmospheric loss can lead to both desiccation and the build-up of abiotic ${\mathrm{O}}_2$ atmospheres that complicate biosignature searches.^17–19

EUV photons are the key drivers for atmospheric mass-loss for three primary reasons: (1) EUV photons are absorbed in the highest (lowest density) layers of the atmosphere, where radiative losses (proportional to density squared) are minor and the heating efficiency is highest, (2) the ionization cross-sections for dominant upper atmospheric atomic species (e.g., H, N, and O) peak at EUV wavelengths, and (3) there are many more EUV photons than x-rays available on all types of stars to drive this heating. In the quiet Sun, the EUV/x-ray photon production ratio is $\sim 90$.²⁰ For optically inactive early M dwarfs, the EUV/X-ray photon ratio is $\sim 40$,²¹ and even for an active late M dwarf like Proxima Cen, the EUV/x-ray photon ratio is $\sim 16$ (Fig. 1).

Fig. 1

(a) Composite spectrum of the nearby planet-hosting M dwarf Proxima Cen, adapted from France et al. (2016)²² and Loyd et al. (2016)²³ The width of the shaded region indicates the approximate uncertainty on the intrinsic flux level. (b) Different reconstructions of the EUV spectrum of Proxima Cen show factors of 3 to 100 flux discrepancies: differential emission measure models in red, blue, and gray (Drake et al. 2020);²⁴ scaling relations based on only FUV data in black (Linsky et al. 2014)²⁵ and only x-ray data in orange (Sanz-Forcada et al. 2011).²⁶ (c) The corresponding atmospheric hydrogen mass lost from an Earth-like planet (in units of Earth oceans; $1\text{Earth ocean}=1.4\times {10}^{24}\mathrm{g}$) from 10 Myr to 4.8 Gyr for high (shaded gray) and low (hatched) EUV histories of typical M dwarfs (Ribas et al. 2017).²⁷

The discovery of rocky planets in the HZs of nearby stars, e.g., Proxima Cen b and the TRAPPIST-1 planets, has motivated new atmospheric mass loss calculations that highlight the need for improved EUV irradiance data to estimate their long-term habitability. Garcia-Sage et al.²⁸ and Airapetian et al.²⁹ presented studies of EUV-driven ion loss from Earth-like planets orbiting M dwarfs. They found mass loss rates several orders of magnitude higher than that of present-day Earth for EUV fluxes 10 to 20 times the current day EUV solar irradiance. Their models indicate that elevated EUV fluxes, augmented by persistent flares or the long pre-main-sequence phase of M dwarfs, could effectively render some Earth-like planets around M dwarfs devoid of an atmosphere, in the absence of internal or external resupply of volatiles. A factor of ten in the uncertainty of incident EUV flux can lead to three orders of magnitude in the uncertainties of ion escape rates,²⁸ fundamentally shifting our predictions for which exoplanets are the best candidates for long-lived habitable atmospheres.

Figure 1 shows the results of analogous thermal atmospheric escape calculations for Proxima Cen b under different plausible EUV radiation strengths and histories. Model calculations yield a factor of $\sim 30$ spread in the total atmospheric mass loss from the planet over time, driven almost entirely by what assumptions are used to estimate the current EUV flux and its evolution earlier in the star’s history.

Despite their importance, the EUV luminosity, flare rate, and evolution of other stars are almost completely unconstrained and can only be empirically determined with direct measurements of the EUV flux and temporal variability of a sample of stars with a range of ages and activity levels. Existing EUV spectra of exoplanet host stars are scarce. The only previous EUV astronomy mission, the extreme ultraviolet explorer (EUVE; Bowyer³⁰ and Malina 1991), obtained spectra of $\sim 15$ cool main sequence stars, including 5 M dwarfs. EUVE’s observations were heavily biased toward the most active stars. The very modest $\le 2{\mathrm{cm}}^2$ peak effective area of the EUVE spectrometers precluded useful spectroscopic observations of stars with more solar-like activity, except for the nearby $\alpha$ Cen system and the F4 subgiant Procyon.^31,32 No observed EUV spectra exist of the optically inactive M dwarfs (defined here as Ca II H and K equivalent widths $<1Å$),^22,33 which will be optimal for atmospheric studies with JWST, 30-m telescopes, and the Origins Space Telescope. The lack of direct EUV data hampers our ability to understand how habitable atmospheres evolve with time and to design truly definitive biosignature searches.

3.2.

EUV Evolution and Flares

Solar EUV emission varies by factors of up to $\sim 100$ on minute timescales due to flares and by factors of $\sim 6$ on year timescales due to the solar cycle.³⁴ Different emission lines and continua vary by different factors on all timescales, whereas connections between the flare energy and frequency distribution of white light flares [such as those observed with $\text{Kepler}/K2$ and Transiting Exoplanet Survey Satellite (TESS)] and UV flares remain poorly constrained by observation and theory.^35–38 Estimates of the EUV response of the Great AD Leo Flare of 1985³⁹ observed with the International Ultraviolet Explorer indicate that the atmospheres of planets orbiting very active stars with frequent flares never achieve a steady state because the timescales for atmospheric recovery are much longer than the time between successive flares.⁴⁰ EUVE provided constraints on flare frequencies for a small sample of young, active stars,⁴¹ however, the lack of direct constraints on EUV flare activity on older stars (e.g., France et al. 2020)⁴² hinders our ability to model the stability of potentially habitable atmospheres over timescales required for the initial development of surface life ($\sim 1.7\mathrm{Gyr}$; Jones and Sleep).⁴³ ESCAPE employs a photon-counting detector to record all observations in a “time-tagging” mode that provides temporal resolution ($<20\mathrm{s}$, limited by photon statistics, the instrumental timing resolution is $<0.1\mathrm{s}$) higher than the characteristic UV flare duration on low-mass stars ($\sim 300\mathrm{s}$; Hawley et al. 2003; Loyd and France 2014; Loyd et al. 2018a).^44–46

Characterization of the EUV flare frequency and spectral variability on stellar evolutionary timescales requires (1) directly collecting EUV spectra over long enough temporal baselines to monitor variability and (2) a broad sample of stars with a range of masses and ages (activity levels, e.g., West et al.).⁴⁷ The required monitoring observations cannot be collected from the ground because of the lack of optical features whose behavior is closely linked to the EUV emission. For example, inactive M dwarfs that show little optical lightcurve variation at the $<1\%$ level can display factors of $\sim 50$ increases in the FUV and x-ray,^22,46 and we expect the behavior to be similar at EUV wavelengths. ESCAPE employs a dedicated monitoring program to sample the flare frequency and energy distribution for individual systems across spectral classes and ages (Sec. 5).

3.3.

Detecting Stellar CMEs

High-energy particles have a major influence on the evolution of exoplanetary atmospheres, both through the actions of stellar winds and impulsive events (e.g., CMEs). While stellar winds play a part in atmospheric escape, CMEs are thought to be the dominant source of long-term instability in planetary atmospheres.^48–50 The influence of CMEs depends strongly on their frequency and ability to break out of coronal magnetic confinement; ESCAPE will measure the frequencies of CMEs on solar-type stars and search for CME breakout from M dwarfs.

While the main source of heat in the upper atmosphere of planets is the EUV flux, Joule heating and particle precipitation heating could dominate the chemical impact of impulsive events (e.g., Segura et al. 2010³⁵ and Tilley et al. 2019).⁵¹ CMEs deliver heat to the upper atmospheres of orbiting planets through charge exchange reactions.⁵² Pickup and sputtering processes can also be major sources of atmospheric loss from direct particle precipitation and stellar wind enhancements due to CMEs. CMEs contribute to atmospheric mass loss on solar system planets^53,54 and to space weather, leading several authors to propose an expanded definition of the HZ that takes into account space weather impacts (e.g., Airapetian et al. 2017 and 2020)^15,29 as more meaningful than the traditional liquid water HZ.

About $\sim 90\%$ of large (X-class) solar flares are associated with CME-like particle eruptions;⁵⁵ however, this connection has not been borne out with recent observations of stellar flares.^56–58 While the EUV irradiance is increased by a large flare rate, CMEs and accelerated particles may have much greater impacts on atmospheric photochemistry and stripping than the flare radiation itself.^35,49

For the Sun, we are able to observe CMEs directly with coronagraphs and in situ measurements. These traditional methods are not feasible for stellar CMEs in the near-term (although see Haisch et al. and Moschou et al.).^59,60 However, other techniques for detecting solar CMEs have been developed. Harra et al.⁶¹ reviewed these techniques and determined that coronal dimming, described below, is the only feature consistently associated with CMEs that can be employed to detect and quantify these events on other stars.

Solar coronal dimming studies have primarily relied upon spatially resolved EUV images to characterize the transient voids left behind in the corona as a CME departs, giving rise to the observed flux dimming (e.g., Sterling and Hudson, Aschwanden et al., Reinard and Biesecker; Dissauer et al. 2018a,b).^62–66 The SDO-EVE, Woods et al. 2012,³⁴ demonstrated that dimming can also be characterized in disk-integrated EUV spectra.^67–70 Similar dimming events have recently been reported in x-ray lightcurves of active stars and an archival EUVE spectrum of the young K star AB Dor.⁷¹ ESCAPE employs the same Sun-as-a-star stellar CME characterization techniques validated on SDO-EVE (Fig. 2) to measure the frequency of CMEs on nearby stars for the first time.

Fig. 2

(a) Observation of coronal dimming in full-disk solar spectra and (b) angularly resolved filter imaging. The spectrally resolved lightcurves from SDO-EVE show flaring lines (Fe XV 284 Å, green) and the corrected Fe IX 171 Å dimming profile in black (Mason et al. 2016).⁶⁸ ESCAPE is sensitive to similar events on F, G, and K dwarfs to a distance of 6 pc. More than 90 candidate events are expected in the DEEP survey (Sec. 5.2).

The dimming light curve also contains information about the kinematics of the CME that produced it. Solar dimming lightcurves have been calibrated by in-situ proton measurements and coronagraphic imaging of CMEs. The depth of the dimming event is directly related to the mass of the ejected CME (the dispersal of the quiescent corona greatly reduces the emission measure, which scales as density squared), whereas the slope of the dimming lightcurve indicates the CME propagation velocity.^68,69 Therefore, observations of coronal dimming events on solar-type stars can provide direct constraints on the physical properties of the CME.

5.1.

Stellar EUV ENvironments Survey

The Stellar EUV ENvironments (SEEN) survey measures the 80 to 1650 Å irradiance and time variability for 200 stars with absolute photometric uncertainty $<40\%$ in the 20 month baseline mission. The photometric accuracy requirement is selected to remove the stellar flux as a dominant source of uncertainty in atmospheric escape calculations.¹⁶ We require the sensitivity and spectral resolution to separate stellar emission lines and measure variability in individual spectral features. These requirements drive the telescope size and grazing incidence design to maintain high throughput at EUV wavelengths (Sec. 6). To illustrate the comparison of a SEEN survey observation with archival data available from EUVE, we show a simulated raw spectrum of Proxima Cen in Fig. 4.

Fig. 4

A simulated ESCAPE SEEN spectrum of Proxima Cen is shown (black). The 77ks EUVE exposure of Proxima Cen (green, offset by 500 counts) resulted in tentative detections of $\sim 5$ stellar emission lines. ESCAPE achieves a line-integrated S/N of $\sim 41$ at Fe IX 171 Å in 35 ksec. Prominent iron ionization states are labeled in red and other species in blue.

About $\sim 20$ flares are required to develop a flare-frequency distribution with $\sim 50\%$ logarithmic rate accuracy (e.g., the rate constant of the flare-frequency power-law distribution). To determine the required SEEN survey duration that effectively constrains the flare rates of F, G, K, and M dwarfs, we estimate the EUV flare rate based on the FUV flare-frequency distribution of Loyd et al. (2018a)⁷⁹ and the FUV-to-EUV scaling relations from France et al. (2018).⁸⁰

The flare frequency distribution is a power law, so more low energy flares are expected than high energy flares. Taking an intermediate activity M dwarf at 10 pc as our fiducial system, we find flares with total Fe IX 171 Å energy $E(\mathrm{Fe}\mathrm{IX})\ge 5\times {10}^{29}$ erg produce a 5-$\sigma$ flare detection with ESCAPE (assuming a typical 5-min M dwarf UV flare duration). This fiducial flare corresponds to a factor of $\sim 15$ brightening, typical for transition region flares observed on low-mass stars.^44,46,79

From the estimated EUV flare rate, we find that in a 35 ksec SEEN observation, we expect to detect 9 flares with $15\times$ brightening $E(\mathrm{Fe}\mathrm{IX})={10}^{29.75}\mathrm{ergs}$; 3 flares with $50\times$ brightening $E(\mathrm{Fe}\mathrm{IX})={10}^{30.25}$; and 1 flare with $100\times$ brightening $E(\mathrm{Fe}\mathrm{IX})={10}^{30.55}$. To assemble statistically significant EUV flare-frequency distributions ($\ge 20$ flares), we will group 3 to 5 stars with similar mass and age properties. Flaring rates and fluxes of young G and K stars are comparable to intermediate activity M stars;^15,81 therefore, we expect young G and K stars to be well represented by the inactive M dwarf calculations presented above.

ESCAPE measures the critical pre-main sequence history of M dwarf EUV luminosities,^82,83 as well as the evolution of the EUV luminosity of F, G, and K stars (e.g., Johnstone et al. 2021; Fig. 5). The SEEN Survey covers a range of ages of each spectral type, spanning pre-main-sequence stars in the TW Hya and $\beta$ Pic moving groups (ages $<30\mathrm{Myr}$), intermediate age clusters such as the AB Dor and the Hyades (hundreds of Myr; Schneider et al.),⁸⁴ and “field” age stars comparable to the Sun (Gyr, based on gyrochronolgy and UV/optical activity indicators). Each spectral type is broken down into approximate “early” (subclass 0 to 4) and “late” (subclass 5 to 9) subgroups, e.g., [early F type stars (F0V to F4V), late-F type stars (F5V to F9V), early G type stars (G0V to G4V), etc.]. The SEEN survey data allow us to assess how similar the EUV outputs of two stars of comparable mass and age are, and importantly, allows us to develop a range of EUV flux estimates for a given set of basic stellar parameters that provides empirical bounds for studies of exoplanets beyond the lifetime of the ESCAPE mission.

Fig. 5

Estimated range of the EUV radiation field on stellar evolutionary timescales, characterized in ESCAPE ’s SEEN survey. The EUV luminosity is the estimated 90 to 360 Å intrinsic luminosity and the symbol sizes represent the integrated S/N of Fe IX in a 35 ksec SEEN observation. Stars with confirmed exoplanets are shown as triangles. This plot shows our current 200 star target list, with resolution in both age and mass (color-coded).

A full 200-star target list that meets these requirements at sufficient S/N is shown in Figs. 5 and 6. About 41 of the baseline stars host known planets (with 85 total planets, 15 of which are in the 200 K to 330 K temperate zone). While this target list satisfies ESCAPE’s science objectives, ongoing results from TESS and Characterising Exoplanets Satellite (CHEOPS) will inform the final target list that will be finalized prior to launch. Finally, it should be noted that exoplanet populations statistics tell us that many of ESCAPE’s “planetless” target stars likely host planets that are yet to be detected (Dressing and Charbonneau; Stark et al. and references therein).^85,86 Future direct-detection missions (e.g., LUVOIR and Habex) will both find and characterize these Earth-like planets around nearby solar-type (F, G, and K) stars. ESCAPE’s SEEN survey ensures that when planets are directly imaged around these stars and spectroscopic characterization begins, the fundamental stellar constraints on the long-term stability of their atmospheres will be in place.

Fig. 6

The ESCAPE target list consists of 41 known exoplanet host stars, and is representative of the underlying known exoplanet population. (a) The known exoplanet population is shown as bolometric flux received against the host star’s effective temperature. (b) The known exoplanet population is shown as distance against stellar effective temperature. Solar system planets are shown as large gray/white circles, exoplanets are shown as small grey circles, and ESCAPE planned targets are blue squares (filled = known exoplanets, empty = no exoplanets yet known). An approximate estimate for the temperate zone (recent-Venus to early Mars; Kopparapu et al. 2013)⁸⁷ is shown as the gray shaded region.

5.2.

Dedicated EUV Eruption Program

The Dedicated EUV Eruption Program (DEEP) survey executes monitoring campaigns of 24 select stars to measure EUV flare frequency distributions and CME rates on individual, high-priority targets in the solar neighborhood.

ESCAPE measures the EUV flare frequency distribution of carefully selected F, G, K, and M dwarf systems on temporal baselines of $\sim 20$ to 40 times longer than the best available datasets from HST. Similarly, there has only been one comparable M dwarf campaign at x-ray wavelengths (0.15 to 15 keV; Kowalski et al. – in prep), and these data do not sample the ${10}^{5-6}\mathrm{K}$ emission that contributes the majority of the stellar EUV flux.⁸⁸ Archival Kepler and TESS observations of white-light flares may also provide connections between the optical and EUV frequencies of superflares (e.g., Loyd et al. 2018b;⁴⁶ Howard et al. 2020).³⁸

The average CME rate of the Sun is $\sim 3.5\mathrm{CME}{\text{day}}^{-1}$ and is $\sim 10\mathrm{CME}{\text{day}}^{-1}$ at solar maximum.⁸⁹ The Sun is less active than other stars of its type (e.g., Reinhold et al. 2020),⁹⁰ so we conservatively estimate a CME rate of $5\mathrm{CME}{\text{day}}^{-1}$ for Sun-like stars. Stellar CME rates are likely well-approximated by a Poisson distribution like the Sun, so the uncertainty on the CME frequency per day is the square root of the number of CMEs observed in a monitoring campaign divided by the monitoring duration. Therefore, observing nine CMEs results in a $3\text{-}\sigma$ constraint on the CME rate. While not every CME is preceded by a dimming event, most are. Veronig et al. find that 97% of CMEs have a corresponding irradiance dimming. However, only $\sim \hspace{0.17em}50\%$ of those will occur on the observable side of the star. Finally, approximately one-third of all solar dimming events meet ESCAPE’s $3\text{-}\sigma$ spectrophotometric detection threshold (dimming depths $\ge 6\%$ in Fe IX 171 Å and Fe X 174 Å). Combining the above numbers with ESCAPE ‘s DEEP survey observing efficiency (78%), the expected detection rate for solar-like CMEs with ESCAPE is approximately $0.63\mathrm{CMEs}/\text{day}$. ESCAPE will monitor each star in the DEEP survey for $\approx 14$ days to provide sufficient temporal baselines to observe nine CMEs/target, resulting in a $3\text{-}\sigma$ determination of the CME rate per star, from which CME-driven atmospheric escape will be estimated.

ESCAPE’s spectral coverage and spectral resolution ($\mathrm{\Delta }\lambda <1.5Å$) are designed to spectrally isolate specific emission lines for temporally resolved lightcurve analysis over a range of coronal ionization states (Fe VIII–Fe XX). ESCAPE has the sensitivity to detect solar-like dimming events (6% max dimming depth in Fe IX 171 Å and Fe X 174 Å, summed over ten 30 minute exposures) to a distance of 6 pc. ESCAPE will survey 10 F, G, and K stars for solar-like CMEs as part of the DEEP survey. For these 10 F, G, and K stars, we expect ESCAPE will detect over 90 solar-like coronal dimming events. Therefore, ESCAPE will likely characterize the frequency of stellar CMEs for the first time. Deeper dimming events or those detectable in broadband spectral binning⁷¹ are even more readily detectable in ESCAPE time-series observations. We note that while dimming events from the hot coronae associated with very active stars may be detectable in the x-rays, EUV spectral coverage is required to sample the lower quiescent coronal emission temperatures typical of the more numerous, lower activity, planet hosting stars.

Magnetic fluxes on active M stars can be factors of 10 to 1000 times higher than global solar magnetic fluxes,^91,92 potentially trapping the charged particle explosions associated with CMEs.⁹³ Simulations by our team indicate that CMEs with energy $>{10}^{34}\mathrm{ergs}$ will escape magnetic confinement from active M dwarfs and be detectable with ESCAPE (Jin et al. – in prep). ESCAPE’s observational capabilities enable us to search for EUV dimming signatures from M dwarfs for the first time, providing empirical constraints on the particle deposition into the atmospheres of orbiting planets.

6.1.

Instrument Subsystems

In the following sections, we describe the key components of the ESCAPE instrument in greater detail.

6.2.

Instrument Performance

The system effective area is a function of the reflection efficiency of the optics ($R$), efficiency of the gratings (${ϵ}_g$), and quantum efficiency of the detector, multiplied by the geometric collecting area of the telescope (${\mathrm{A}}_{\mathrm{geo}}$):

The total $A_{\mathrm{eff}}$ of ESCAPE (Fig. 8) is the sum of all diffraction orders, as well as the sum of the two GI channels for where the wavelength range overlaps. The $A_{\mathrm{eff}}$ of the NI channels (G70 and G140) is calculated with added terms for the fold mirror, ${\mathrm{CaF}}_2$ filter (G140), and zero order reflection off of the corresponding GI grating.

Fig. 8

A comparison of the ESCAPE effective area (solid black line) compared with previous astrophysics missions covering the 100 to 912 Å EUV bandpass. ESCAPE intentionally introduces a gap in spectral coverage between $\sim 550$ and 600 Å to avoid strong O II and He I geocoronal emission.

The point-spread-function (PSF) is the sum of contributions from the intrinsic grating distortion, optical mount stresses, 1-sigma pointing jitter, maximum allowable mirror fabrication errors and maximum allowable detector resolution. This PSF produces a predicted RMS spectral resolution performance at 171 Å for the grazing incidence channels of 0.91 Å in G20 and 0.66 Å in G40 (second order), and normal incidence performance of 3.6 Å in G70 and 6.1 Å in G140 (see Table 1), averaged over the normal incidence mode bandpasses. The raytraced PSF spot for the G20 at 171 Å is shown in Fig. 9.

Fig. 9

Projected PSF for the G20 mode at 171 Å, sampled at the detector digital pixel scale ($\sim 14\mu \mathrm{m}\times 14\mu \mathrm{m}$).

6.3.

Simulated Data Example

We carried out science performance calculations using a conservative spot size corresponding to the baseline requirement of 1.5 Å spectral resolution at 171 Å, lower than the projected instrument performance. A simulated two-dimensional detector image of a SEEN survey target is shown in Fig. 10(a). ESCAPE’s targets are cool stars, dominated by coronal and chromospheric line emission. Therefore, the order overlap associated with continuum sources that was an issue for EUVE is mitigated for ESCAPE and wavelength identification can be accomplished without wavelength blocking filters. For ESCAPE, the wavelength definition can be cleanly inferred from comparison of one-dimensional (1D) spectral extractions with solar or synthetic stellar spectra (as illustrated by the minimal spectral order overlap for solar-type stars shown in Fig. 10, center panels). More sophisticated forward modeling techniques, used on grating spectra at x-ray wavelengths (see, e.g., Brinkman et al.),⁹⁴ will be developed to identify spectral features by their location in the two-dimensional (2D) spectrogram.

Fig. 10

A schematic representation of the ESCAPE science data flow for an inactive solar type star at a distance of 7 pc, with a typical SEEN exposure time of 35 ksec. (a) Level 3a science products contain the ($\lambda$, $y$, $t$) photon lists, including order identifications. The oval features are He II 304 Å airglow, which are spectrally isolated on the detector and do not impact the majority of the science spectrum. (b) 1D spectra are extracted and different wavelengths are readily identified in the G20 and G40 channels. (c) Order-sorted 1D spectra are combined with interstellar attenuation corrections and differential emission measure models to produce full extreme ultraviolet irradiance spectra.

Spectral data are order-sorted by three methods: (1) laboratory spectra obtained during instrument integration and testing provide a map of spectral detector locations as a function of input wavelength for known calibration gases, (2) on-orbit checkout calibration observations of cool stars with EUVE observations will be used to validate forward modeling analyses of coronal models; pre-flight simulations show clear delineation of different wavelengths in 1D spectral traces for G20 and G40 channels (Fig. 10, center), and (3) order-sorting filters (manufactured by Luxel) isolating short-, medium-, and long-wavelength EUV bands are well-established in solar missions and will be used to validate the forward modeling order sorting analysis as well as reject soft x-ray continuum leaking into the ESCAPE band for stars with high-temperature coronae.

An Al+Mg filter isolates the first order G40 spectrum (and supports the calculation of a predictive DEM model to guide the extraction). The G40 Open – Al+Mg difference spectra are cross-correlated with the $\mathrm{Zr}+{\mathrm{MoSi}}_2$ G20 spectra to isolate the first and second order G20 and G40 spectra. An indium filter tests for short-wavelength bremsstrahlung leaking into the ESCAPE band. Filters are only used for initial wavelength/order calibration activities; an open aperture position is used for all SEEN and DEEP survey observations.

Figure 10 shows the typical data flow for ESCAPE observations. The Level 2 products contain the instrument data organized by observations, photon lists at full detector resolution, with all of the detector related corrections applied (application of ground-based flat-field, thermal and geometric distortion maps, and 2D walk correction as necessary; e.g., COS Data Handbook 2015; France et al. 2011¹¹¹). The photon-list data are time-tagged to 100 msec accuracy.

Two Level 3 (L3) data products are generated: The L3a products preserve the full time-tag photon list from Level 2, with the addition of corresponding grating order probability (the fraction contributed by each order to the best-fit forward model spectrum at a given detector position) and wavelength for the four expected orders. Wavelength solutions are provided to the end-user for the creation of lightcurves in individual spectral features (as is done for HST-COS). Level 3a science products contain the ($\lambda$, $y$, and $t$) photon lists that are required for stellar flare and CME characterization [Fig. 10(a)]. L3b contain the fully calibrated, one dimensional spectrum in ergs ${\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}{Å}^{-1}$ as a function of wavelength (in Å) for each observation (both in individual orders and a final coadded spectrum). L3b data are combined with interstellar attenuation corrections and a DEM model to produce full extreme ultraviolet irradiance spectra [Fig. 10(b)]. The ESCAPE data will be hosted at the Mikulski Archive for Space Telescopes (MAST) archive.

Level 3a and 3b data products will be produced for individual exposures and exposure time integrated for each ESCAPE target to facilitate dissemination and use by the astronomical community. Level 3b science products contain the full 1D EUV and FUV stellar spectra that are required to measure and reconstruct the EUV irradiance incident on orbiting planets. Figure 4 shows a simulated SEEN observation (extracted counts spectrum) of the nearby planet-hosting M dwarf Proxima Cen, compared to an archival EUVE spectrum of the same source. Figure 11 shows a Level 3b data product of a simulated DEEP observation (the flux spectrum) of Proxima Cen and the predicted signal-to-noise ratio as a function of wavelength that includes the influence of local ISM absorption on the ESCAPE observations.

Fig. 11

(a) Simulated observation of the full EUV spectrum of Proxima Cen comparing the intrinsic (blue) and ISM-absorbed ESCAPE data (orange). (b) Signal-to-noise per 1.5 Å resolution element of a DEEP Survey observation of Proxima Cen.