JWST MIRI MRS Images of Disk Winds, Water, and CO in an Edge-on Protoplanetary Disk

We present JWST MIRI MRS observations of the edge-on protoplanetary disk around the young subsolar-mass star Tau 042021, acquired as part of the Cycle 1 GO program “Mapping Inclined Disk Astrochemical Signatures.” These data resolve the mid-IR spatial distributions of H2, revealing X-shaped emission extending to ∼200 au above the disk midplane with a semiopening angle of 35° ± 5°. We do not velocity-resolve the gas in the spectral images, but the measured semiopening angle of the H2 is consistent with a magnetohydrodynamic wind origin. A collimated, bipolar jet is seen in forbidden emission lines from [Ne ii], [Ne iii], [Ni ii], [Fe ii], [Ar ii], and [S iii]. Extended H2O and CO emission lines are also detected, reaching diameters of ∼90 and 190 au, respectively. Hot molecular emission is not expected at such radii, and we interpret its extended spatial distribution as scattering of inner disk molecular emission by dust grains in the outer disk surface. H i recombination lines, characteristic of inner disk accretion shocks, are similarly extended and are likely also scattered light from the innermost star–disk interface. Finally, we detect extended polycyclic aromatic hydrocarbon (PAH) emission at 11.3 μm cospatial with the scattered-light continuum, making this the first low-mass T Tauri star around which extended PAHs have been confirmed, to our knowledge. MIRI MRS line images of edge-on disks provide an unprecedented window into the outflow, accretion, and scattering processes within protoplanetary disks, allowing us to constrain the disk lifetimes and accretion and mass-loss mechanisms.


Introduction
Planets form within circumstellar disks of gas, dust, and ice, where they must accumulate material before it is removed via outflow mechanisms and accretion onto the central stars.Observations suggest that this process takes at most a few Myr, as determined by the age at which the submillimeter dust and gas content in disks decreases significantly (see, e.g., Barenfeld et al. 2016;Pascucci et al. 2016;Simon et al. 2019).Models of viscously evolving disks predict a corresponding decline in mass accretion rates, as the gas reservoir is also depleted over time (Lynden-Bell & Pringle 1974;Hartmann et al. 1998).However, this picture is complicated by significant scatter in the observed relationship between mass accretion rates ( ) M acc  and disk dust masses (M disk ;Hartmann et al. 1998;Betti et al. 2023;Manara et al. 2023), which indicates that additional physical mechanisms are required to drive mass loss and set the maximum lifetimes of protoplanetary disks (Pascucci et al. 2023).
Recent work demonstrates that the relationship between M acc  and M disk , including the scatter, can be reproduced by models where magnetohydrodynamic (MHD) winds are the primary drivers of angular momentum transfer that enables ongoing accretion (Tabone et al. 2022b;Somigliana et al. 2023).In this framework, a disk population begins with a distribution of initial disk sizes and viscosities (Tabone et al. 2021).A magnetized disk wind, removing mass at approximately the same rate as accretion, can then drive disk dispersal over a timescale consistent with observations (Blandford & Payne 1982;Cabrit et al. 1999;Bai & Stone 2013;Lesur 2021).
Although MHD wind models can successfully reproduce the decline in M acc  and M disk over time, an unambiguous distinction between outflow-launching mechanisms in the inner disk remains elusive (jets, MHD, and photoevaporative winds; see, e.g., Pascucci et al. 2023 for a recent review).Outflow signatures are readily identified in high-resolution spectroscopy, which reveals blueshifted emission and absorption lines from atomic transitions, including [O I], [Ne II], [N II], [S II], He I, and C II (Edwards et al. 1987;Kwan & Tademaru 1995;Pascucci & Sterzik 2009;Fang et al. 2018Fang et al. , 2023a;;Banzatti et al. 2019;Haffert et al. 2020;Xu et al. 2021;Campbell-White et al. 2023).The central velocities and FWHM of the line profiles, which can be interpreted as average outflow velocities and radial locations within a Keplerian disk, are typically used to distinguish between an origin in photoevaporative and/or MHD winds.For example, a more compact emitting region (r ∼ 0.1-0.2au) is assumed to be characteristic of an MHD wind (Campbell-White et al. 2023).Particularly the line flux ratios of [Ne II] 12.814 μm and [O I] 6300 Å in low-velocity photoevaporative winds increase as emission from the warm inner disk dust declines (Pascucci et al. 2020).A similar correlation is detected between the inner disk dust content and the morphology of accretion-generated Lyα (Arulanantham et al. 2023) and Hα emission lines (Fang et al. 2023a), indicating that the relative contributions of MHD and photoevaporative winds and accretion to the total mass-loss rate may evolve over time.
Searches for spatially extended gas from protoplanetary disk winds have recently detected [O I] and Lyα emission from the face-on system TW Hya (Fang et al. 2023b;Chang et al. 2024).The [O I] morphology is reproduced by models of a radially confined MHD wind launched from within the dust sublimation zone, while the Lyα distribution is consistent with resonant scattering from the accretion shocks through H I within both a wind and the disk itself.Spectroastrometry of [O I] emission in RU Lupi and AS 205 N (Whelan et al. 2021) reveals a more extended emitting region consistent with the superposition of a wide-angled MHD disk wind (0 km s −1 < |v| < 40 km s −1 ) and an accelerating jet (90 km s −1 < |v| < 230 km s −1 ).Spatially extended near-infrared emission from rovibrational H 2 lines has also been identified from T Tauri systems (Eislöffel et al. 2000;Duchêne et al. 2005;Beck et al. 2008;Beck & Bary 2019), as expected from high-resolution 1D spectroscopy that showed blueshifted emission line profiles (Beckwith et al. 1978;Bary et al. 2003;Gangi et al. 2020).The emission lines originate in gas with temperatures between T ∼ 1800 and 2300 K and yield line flux ratios consistent with shock excitation (Eislöffel et al. 2000;Beck et al. 2008).Mid-infrared emission from rotational quadrupole transitions of H 2 has been detected at temperatures between T ∼ 100 and 200 K, showing line profiles consistent with extended emission (Thi et al. 2001;Bitner et al. 2008).Still, it remains observationally challenging to distinguish whether MHD or photoevaporative winds are responsible for driving the detected outflows, as the wind launching radii and velocity structure remain spatially and spectrally unresolved at the distance to nearby star-forming regions (see, e.g., Beck et al. 2008;Coleman et al. 2024).Clues may come from spatially resolved observations of [Ne II], where the emission line luminosity increases as dust cavities are cleared in the disk and the contribution from photoevaporative winds becomes more prominent (Pascucci et al. 2020).
In this paper, we report the detection of spatially extended rotational H 2 , H 2 O, and CO emission from an edge-on protoplanetary disk with JWST MIRI Medium Resolution Spectroscopy (MIRI/MRS) using integral field unit (IFU) observations.We describe the target and observations and our method for constructing line and continuum images in Section 2. In Section 3, we compare the morphology of the molecular lines to emission lines from forbidden atomic transitions and H I Humphreys α emission, which allows us to consider winds, jets, and accretion as drivers of mass loss within the same system.We also compare the line images to the spatial distribution of mid-infrared scattered light to constrain the origin of the extended molecular gas emission.The kinematics of the extended emission are discussed in Section 4, and we conclude in Section 5.

Description of Observing Program
The target object, 2MASS J04202144+2813491 (hereafter Tau 042021; R.A. 04:20:21.44,decl.+28:13:49.17),surrounds a low-mass star (M1, 0.27 M e ; Simon et al. 2019) at a distance of ∼140 pc surrounded by a 5″ edge-on disk (>85°) first seen at optical wavelengths with the Canada-France-Hawaii Telescope (CFHT; Luhman et al. 2009) and later imaged with the Advanced Camera for Surveys (ACS) on board the Hubble Space Telescope (HST; Stapelfeldt et al. 2014;Villenave et al. 2020).It is also one of the few edge-on disks for which both the radial and vertical structure have been resolved in Atacama Large Millimeter/submillimeter Array (ALMA) observations (Band 4, 3 67 × 0 2; Villenave et al. 2020Villenave et al. , 2023)).New JWST images of the system acquired with NIRCam and MIRI reveal a well-mixed population of 10 μm dust grains, which are coupled to the gas in surface layers down to the midplane; the same images also show an X shape in the F770W and F1280W filters, which may be associated with a disk wind (Duchêne et al. 2024).A bipolar, collimated jet is detected as well, using the F606W filter on HST-ACS (Duchêne et al. 2014) and JWST MIRI imaging at 12.8 μm (Duchêne et al. 2024).We selected Tau 042021 for the MIDAS program on the basis of its 6 and 6.8 μm ice features, along with the strong H 2 and [Ne II] 12.8 μm gas emission lines and broad 11.3 μm polycyclic aromatic hydrocarbon (PAH) emission that were visible in its Spitzer InfraRed Spectrograph (IRS) spectrum (PI: J. R. Houck; PID: 50053; see, e.g., Furlan et al. 2011;Sturm et al. 2023;Duchêne et al. 2024).
Our JWST MIRI MRS observations were obtained on 2023 February 13 as part of PID 1751 (PI: M. McClure).Since the target is extended and has HST coordinates, we did not use target acquisition.Although the disk diameter is 5″ at optical wavelengths, which is larger than the MIRI detector between 4.9 and 7.65 μm, we modeled its spatial extent using radiative transfer models fit to the HST image and Spitzer spectrum and determined that the mid-infrared scattered-light image would fit within MIRI's field of view (FOV).We used a four-point dither pattern optimized for an extended source, with each dither exposure consisting of one integration of 69 groups, read using the FASTR1 readout pattern, for total on-source exposure times of 6594 s in each of the MIRI channels.This integration time was selected to get a signal-to-noise ratio (S/N) of 10 per spaxel on the central 3 × 5 spaxel region of our simulated disk model, which would enable detection of CH 4 and other weak ices after rebinning.Since the source is extended and large, a dedicated sky background was obtained with dithers of the same number of groups but a two-point dither pattern to reduce the nonscience time.The final 3D IFU cube was produced using version 11.17.1 of the Calibration References Data System (CRDS) file selection software in the JWST Science Calibration Pipeline (CRDS context "jwst_1174.pmap") with stage 3 processing utilized to combine data from all dither positions (Bushouse et al. 2023).The 1D spectra were extracted from the eastern and western lobes of the disk and calibrated using the JDISCS extractor (Pontoppidan et al. 2024).
The spectral resolution of the MRS changes as a function of wavelength.While in-flight observations during the instrument commissioning were used to derive resolving powers between R ∼ 1500 and 3500 (Jones et al. 2023), calibration against highresolution ground-based spectroscopy of the protoplanetary disk system FZ Tau is consistent with R ∼ 2000-3000 (Pontoppidan et al. 2024).Using the coefficients for each subband provided in Pontoppidan et al. (2024), we expect Δv ∼ 106 km s −1 at 5 μm and Δv ∼ 123 km s −1 at 28 μm.

Construction of Line and Continuum Images
To explore the spatial extents of the gas emission lines, we produced continuum-subtracted spectral images for the detected transitions.The 2D local dust continuum was measured as the median surface brightness per spaxel in frames selected from emission-line-free wavelengths along the spectral axis (see, e.g., Pontoppidan et al. 2024;Worthen et al. 2024).Approximately 10 frames were used to estimate the continuum, spanning ±0.007 μm on either side of the emission line.This background image was subtracted from each of the four to five frames containing the line emission, which were then stacked to produce the final image.We note that the target was not centered on the detector during the observations, requiring an additional correction to shift the disk to the zero position in the line images presented here.Since the MIRI MRS IFU cubes produced by the JWST calibration pipeline are registered to the World Coordinate System, we determined the correction as the difference between the R.A. and decl. of Tau 042021 and the image center (1-3 pixels, depending on the channel).We do not resample the images, so any remaining deviations from the zero-point are likely due to the large MIRI MRS spaxel sizes.
For both the CO and H 2 O line series, all detected features were coadded to produce the final line surface brightness maps.The heights vertical to the disk and diameters along the midplane of each emitting region were measured as the area within which the surface brightness is greater than 40% ± 2.5% of the peak surface brightness in the spectral images (see Table 1).The 40% threshold was chosen instead of an FWHM measurement, as this is the limit where emission from both sides of the disk can be distinguished from the residual background in the line images (see, e.g., Beck et al. 2008).The uncertainties are derived from the surface brightness limit where the emitting region shifts into the next adjacent spaxels.We discuss the observed gas-phase morphologies in the following sections.

Results
In the MIRI MRS spectrum of Tau 042021, we detect the mid-infrared dust continuum and gas-phase emission lines from H 2 and [Ne II] that were first observed in the Spitzer spectrum (see Figure 1; Duchêne et al. 2024).Additionally, we detect the 11.3 μm PAH band and gas-phase emission lines from CO (rovibrational), H 2 O (rotational and rovibrational), OH (rotational), H  1.
The 3D spectral cube from Tau 042021 reveals spatially extended emission at wavelengths corresponding to the atomic and molecular gas-phase transitions detected in the 1D spectra (see Figures 2 and 4).At the target distance of d ∼ 140 pc, the pixel scales of the MIRI IFUs increase from ∼27 au (0 196) in channels 1 and 2 to ∼38 au (0 273) in channel 4 (Argyriou et al. 2023).Molecular gas in disks is not expected to produce strong mid-infrared emission lines at such large radial separations from the host stars (see, e.g., Salyk et al. 2011).However, recent MIRI MRS line images of FZ Tau revealed a spatially extended ring of H 2 emission (Pontoppidan et al. 2024).A ring of [Ar II] emission was also detected in the Beta Pictoris system, spatially coincident with its dusty debris disk (Worthen et al. 2024).In the following sections, we present the continuum-subtracted line images of atomic and molecular lines in Tau 042021 and discuss the observed gas-phase morphologies.We find that the western nebula is brighter than the eastern nebula in all tracers reported in Tables 1 and 2, as expected if the western nebula is tilted slightly toward the observer (Simon et al. 2019;Villenave et al. 2020) or if the brightness of the eastern nebula is variable (Luhman et al. 2009).

Scattered and Thermal Dust Continuum Emission
We observe a strong difference between the mid-infrared and millimeter emission of the disk.The scattered-light lobes in Tau 042021 are similar to those seen with the I-band filter on MegaPrime/MegaCam at the CFHT (λ 0 = 0.776 μm; Luhman et al. 2009) and the F606W (λ 0 = 0.5888 μm) and F814W (λ 0 = 0.8115 μm) filters of the ACS on board the HST (Stapelfeldt et al. 2014) but require the presence of 10 μm grains at high disk altitudes (Duchêne et al. 2024).Conversely, the millimeter emission traces thermal emission from cold, large grains in the disk midplane.Each panel in Figure 2 shows the 12 μm dust continuum from Tau 042021, drawn as white contours over the images.The upper left panel depicts the ALMA Band 7 (0.89 mm) emission from Villenave et al. (2020).

Spatially Resolved Mid-IR H 2 Emission in a Protoplanetary Disk
We detect eight H 2 emission lines from the rotational ladder that fall within the MIRI MRS spectral range (see Table 1).The upper right panel of Figure 2 shows the image of the H 2 S(2) (v = 0-0) emission line at 12.278 μm, compared to the scattered-light dust continuum at 12.2 μm.We observe H 2 emission across the entire radial extent of the dust disk,  spanning a diameter of 220 7 8 -+ au in the western nebula.The H 2 S(2) emission is significantly more vertically extended than the dust scattered light, which only reaches 95 3 4 -+ au above the disk midplane.By contrast, the H 2 S(2) emerging from the western nebula extends to 171 ± 7 au.
The vertically extended H 2 S(2) emission forms an X-shaped distribution, centered on the darker disk midplane.This is consistent with the JWST NIRCam and MIRI images of Tau 042021, which revealed an X-shaped structure in the 7.7 and 12.8 μm filters that was attributed to a disk wind (Duchêne et al. 2024).We measured the semiopening angle of each half of the X in the MIRI MRS line images relative to the vertical disk axis (aperture position angle of 81.°896), finding θ = 35°± 5°.The angle with respect to the disk midplane is then θ ∼ 55°, which is steeper than the 36 • angle between the midplane and the X in the 7.7 μm MIRI image (Duchêne et al. 2024).A similar "onion-like" wind structure has previously been reported for DG Tau B, where velocity-resolved images show increasingly wide opening angles in emission lines tracing larger launching radii (Bacciotti et al. 2000;Lavalley-Fouquet et al. 2000;de Valon et al. 2020;Pascucci et al. 2023).This includes the 45°semiopening angle of the wide molecular wind traced by rovibrational H 2 v = 1 − 0 S(1) emission (Beck et al. 2008;Agra-Amboage et al. 2014) and the 90°cone measured from UV-fluorescent H 2 emission in the same system (Schneider et al. 2013).
The 7.7 μm filter on MIRI spans wavelengths between 6.581 and 8.687 μm (Wright et al. 2023), which includes the H 2 S(5) and H 2 S(4) transitions detected in our MIRI MRS spectrum (6.909 and 8.025 μm, respectively).In the line images, the H 2 S(5), S(6), S(7), and S(8) emission lines trace an identical spatial region as the dust at 5 μm, while the H 2 S(1), S(3), and S(4) emission lines are also vertically extended.It is likely then that the X-shaped feature imaged at 7.7 μm by Duchêne et al. (2024) includes a superposition of the vertically extended H 2 S(4) and the less extended H 2 S(5) emission.We find that the sizes of both nebuluae decrease monotonically as a function of wavelength, with the H 2 S(7) emission line at 5.511 μm showing a height of just 94 4 5 -+ au and a diameter of 177 ± 4 au across the brighter, western lobe of the disk.This trend cannot be attributed to the increase in PSF size toward longer wavelengths, since the disk extends well beyond the average PSF FWHM at the corresponding wavelengths listed in Table 1.The spatially extended emitting regions we measure are also consistent with the 1D spectroscopy results of Richter et al. (2002), who reported nondetections of H 2 S(1)-S(8) rotational transitions within the small 2″ × 2″ aperture on the Texas Echelon Cross Echelle Spectrograph in a small sample of protoplanetary disks, despite clear detections with the larger aperture on the Infrared Space Observatory that could better capture extended emission (Thi et al. 2001).
We note that the H 2 emission lines in which the extended emission is most prominent, S(1) and S(2) at 17.036 and 12.278 μm, respectively, are both observed at wavelengths longer than the limit below which the MIRI cruciform artifact appears (λ < 10-12 μm; Gáspár et al. 2021).The feature is a result of lower quantum efficiency at shorter wavelengths in the MIRI detectors, which leaves the remaining photons to scatter internally.Since the extended H 2 emission is observed at longer wavelengths, where the quantum efficiency of the detectors is higher, the images are not contaminated by the artifact.

Bipolar Jets Detected in Forbidden Emission
We report the detection of spatially extended [Ne II] (12.814 μm) emission from Tau 042021 (see Figure 2).Emission lines from [Ne III] (15.555 μm), [Fe II] (5.34, 17.936 μm), [Ni II] (6.636, 10.682, 12.729, 18.241 μm), [Ar II] (6.985 μm), and [S III] (18.713 μm) are also identified and spatially resolved, with median emitting regions between ∼67 and 75 au wide and ∼150 au long (see Figure 1 and Table 1).Like the H 2 S(2) emission shown in Figure 2, the [Ne II] emission extends beyond the vertical height of both dust reflection nebulae observed at 12 μm.However, the [Ne II] appears in four distinct knots, with a much narrower radial extent than the H 2 in the knots that coincide with the dust continuum.The distribution is consistent with emission from a bipolar jet (see, e.g., Dougados et al. 2000), which was discovered in the earlier HST and JWST images of Tau 042021 (Duchêne et al. 2014(Duchêne et al. , 2024)).High-resolution spectroscopic studies of mid-IR [Ne II] emission typically detect either a high-velocity component (HVC) associated with collimated jets or a low-velocity component (LVC) tracing either MHD or photoevaporative disk winds in the 1D emission line profiles (Herczeg et al. 2007;Güdel et al. 2010;Pascucci et al. 2011;Alexander et al. 2014;Pascucci et al. 2020) and 2D line images (Bajaj et al. 2024).Although the MIRI MRS observations do not have the Images have been scaled using an asinh stretch to maximize the contrast between the background noise and the high-S/N emission from the source.Beige circles in the bottom left corner of each panel represent the average theoretical FWHM of the PSF, calculated from the relationship reported in Law et al. (2023).The observed spatial distributions are consistent with micron-sized grains in the eastern and western reflection nebulae (12 μm continuum), a disk wind (H 2 S(2)), a bipolar jet ([Ne II]), and scattered light from accretion shocks or from the base of the jet (H I).We note that the position angle of the jet is tilted by ∼2°. 5 relative to the disk.spectral resolution to distinguish HVCs or LVCs, the X shape observed with the F1280W broadband filter on MIRI (Duchêne et al. 2024) is consistent with the H 2 S(2) emission that we observe and not with the [Ne II] jet.We do not spatially resolve changes in the jet width as a function of distance from the star, which places an upper limit of <3°on the jet opening angle (see, e.g., Narang et al. 2024).Figure 2 shows that the position angle of the [Ne II] jet appears slightly offset from the position angle of the submillimeter disk (PA = −16°; Villenave et al. 2020), and we measure a shift of ∼2°. 5. We discuss possible sources for the misalignment in Section 4.2.

Detection of Spatially Extended PAH Emission
The PAH emission feature seen in the Spitzer spectrum and detected in Figure 1 is also extended with approximately the same distribution as the scattered dust continuum (see Figure 3).The IFU imaging proves that the PAH emission is absent in the background and localized to the disk.With a spectral type of M1 ± 2 (Luhman et al. 2009), this may be the first detection of PAH emission around a low-mass T Tauri star (M-K spectral type) where the PAH emission is intrinsic to the disk itself rather than due to external irradiation (Geers et al. 2006).It may be that the unique geometry of this system, which blocks the thermal emission from the inner disk, allows the relatively weaker PAH emission around T Tauri stars to be detected at mid-infrared wavelengths.The spatial extension of the PAH feature is robust; however, its location at the end of channel 2 means that the red continuum is poorly defined.Therefore, we leave the analysis of this feature to future work.

Scattering of Inner Disk CO and H 2 O Emission
Spatially resolved emission from CO rovibrational lines between 4.9 and 5.3 μm is detected in the MIRI MRS cube, corresponding to the transitions v = 1 − 0 P(25)-P(40).Figure 4 shows the combined emission from these features after subtraction of the local continuum, revealing a span of 187 ± 6 au in diameter and 94 ± 3 au in height above the disk midplane for the western lobe of the disk.The fainter, eastern lobe has a diameter of 150 20 22 -+ au and a height of 73 13 5 -+ au.By contrast, spectroastrometric observations of the rovibrational CO lines were consistent with more compact distributions of hot gas, with local thermodynamic equilibrium (LTE) temperatures of 100-1000 K and emitting areas inside r < 4 au (Pontoppidan et al. 2011;Brown et al. 2013).The emitting region we observe with MIRI in this edge-on disk is significantly larger than expected for lines originating in gas within the 100-1000 K temperature range, indicating that the extended emission is more likely scattered light (see, e.g., Pontoppidan et al. 2002).
Numerous emission lines from the rotational and rovibrational bending mode transitions of H 2 O (5.65-6.19μm; see Table 1) are also readily detected in the MIRI spectral cube.Figure 4 shows the stacked images of rovibrational bending mode features detected within the medium subband of channel 1.The H 2 O distributions extend far beyond the emitting  Geers et al. 2006Geers et al. , 2007)), it is rare to detect the band within disks around M-type stars.We note that the continuum on the red side of the band is beyond the edge of the channel 2 detector, making it challenging to extract the total surface brightness profile.The image shown here is from a single frame where the feature peaks in emission.
regions derived from slab model fits to mid-infrared spectra from less inclined disks (r < 1 au; see, e.g., Salyk et al. 2011;Banzatti et al. 2023;Gasman et al. 2023;Perotti et al. 2023;Muñoz-Romero et al. 2024).The image of the H 2 O rotational lines in Tau 042021 shows a diameter of 113 12 13 -+ au and a vertical extent of 70 10 9 -+ au in the western lobe of the disk.In channel 1, the rovibrational lines between 5.6 and 5.9 μm span a disk diameter of 90 ± 6 au in the bright lobe and a vertical height of 61 ± 3 au.However, the larger pixel scale in channel 3 makes it challenging to directly compare the spatial distributions of the rotational and bending mode emission lines, and the difference in emitting areas reported here is not statistically significant.

Hydrogen Recombination Lines Emerging from Accretion Shocks
The bottom right panel of Figure 2 shows the 2D spectral image of the H I (7-6) Huα emission line at 12.372 μm.We also identify H I Pfα (7.456 μm) and Huβ (7.503 μm) emission lines in the 1D spectrum, which were tentatively detected in the Spitzer-IRS spectrum (Duchêne et al. 2024).However, the peak surface brightnesses from the Pfα and Huβ emission lines (∼5 × 10 −20 W m −2 arcsec −2 ) are close to the background noise level in the continuum-subtracted images, making it challenging to measure the spatial extent of the emitting regions.We constrain the stronger H I (7-6) emission to a diameter of 113 ± 13 au and a height of 92 ± 7 au in the brighter, western nebula.The fainter, eastern nebula has a diameter of 121 ± 3 au and a height of 134 ± 2 au.
Infrared H I recombination lines are typically correlated with accretion luminosity in class I (see, e.g., Fiorellino et al. 2021) and class II sources (see, e.g., Pascucci et al. 2007;Salyk et al. 2013;Rigliaco et al. 2015;Alcalá et al. 2017), implying that the emission originates within the accretion shocks rather than the disk surface.In particular, the Huα transition was detected in 46/114 disks observed with Spitzer (40%), where it was found to trace H I number densities between 10 10 and 10 11 cm −3 (Rigliaco et al. 2015).We measure the integrated flux in the H I (7-6) emission line from Tau 042021, finding F = 2.6 × 10 −20 W m −2 in the western lobe and 1.9 × 10 −20 W m −2 on the eastern side of the disk.However, Figure 2 shows that much of the intrinsic emission line flux is likely scattered out of the line of sight.
In a less inclined disk, the infrared spectral energy distribution (SED) is relatively flat between ∼12 and 100 μm (see, e.g., Dullemond et al. 2007).The SED from Tau 042021 shows that the flux increases from λF λ ∼ 10 −16 W m −2 at ∼12 μm to λF λ ∼ 1 × 10 −14 W m −2 at ∼100 μm (Duchêne et al. 2024) as the contribution from thermal emission becomes visible in the edge-on disk.If the decrease is predominantly an inclination effect, the "true" flux at ∼12 μm is likely also closer to 10 −14 W m −2 .We include this flux ratio of ∼100 as a scaling factor to correct the H I Huα emission line flux for scattering and use the linear relationship between the H I and accretion luminosities (Rigliaco et al. 2015) to derive L acc = 7.9 × 10 −5 L e for Tau 042021.The accretion rate can then be estimated as et al. 1998), assuming R in = 5 R * (Rigliaco et al. 2015), R * = 1.7 R e , and M * = 0.27 M e (Simon et al. 2019).This value is likely a significant underprediction, and we discuss the implications further in Section 4.2.

Extracting the Molecular Gas Thermal Structure
To explore whether the radially extended H 2 , CO, and H 2 O emission is consistent with a distribution of scattered light, we used the spectools_ir package (Salyk 2022) to derive the temperature, column density, and emitting area of the warm gas.spectools_ir produces slab models in LTE, which we fit to the 1D spectrum of Tau 042021 after using the publicly available ctool routine (Pontoppidan et al. 2024) to subtract the continuum.We fit each molecule individually, as the H 2 and CO emission lines are well separated from both the rotational and rovibrational H 2 O transitions.
Table 3 reports the best-fit slab model parameters for all three molecules.We find that the CO spectrum is consistent with T = 1500 K and N log 17.5 = cm −2 , demonstrating that the temperature required to produce the J = 25-40 emission lines is similar to what is observed in less inclined disks (Salyk et al. 2011;Brown et al. 2013).A similar effect is seen in the H 2 O emission lines, which require T = 850 K and

N log
15.0 = cm −2 .Again, this is roughly consistent with JWST MIRI MRS observations of water in less inclined protoplanetary disks, which show temperatures between ∼600 and 900 K (Banzatti et al. 2023;Gasman et al. 2023;Grant et al. 2023;Kóspál et al. 2023;Perotti et al. 2023;Muñoz-Romero et al. 2024).We note that Tau 042021 does not show an excess cool H 2 O component, which has been associated with pebble drift in compact disks (Banzatti et al. 2023).Rather than originating at large radial and vertical separations, the extended emission we observe is likely scattered along the disk surface by large dust grains, an effect uniquely detected in edge-on systems.
In the case of H 2 , we find that two temperature components are required to reproduce the observed spectrum.The first, with T = 400 K and N log 19.4 = cm −2 , captures the bulk of the emission from the S(1)-S(4) emission lines, which are vertically extended in the line images.However, this model underpredicts the S(5)-S(8) emission line fluxes, which are better reproduced when an additional gas population with T = 1500 K and N log 20.1 = cm −2 is included.The hotter temperature is consistent with the CO within the inner disk, while the cooler temperature is closer to the 100 K wind temperature assumed for the mid-infrared H 2 emission lines measured by Thi et al. (2001).However, we note that the emitting area we retrieve for the 1500 K H 2 (r = 5 au) is significantly larger than the intrinsic emitting area of the CO (r = 0.01 au).

No Velocity Structure Detected in Spatially Extended
Emission Lines We explore the velocity structure in the H 2 and atomic emission lines using both the 1D MIRI MRS spectrum and the spectral images.We fit each of the emission lines listed in Table 1 with a model of a Gaussian profile, superimposed on a linear continuum extending ±0.02 μm from the central wavelength of the transition.The free parameters are the amplitude, velocity centroid, and width of the Gaussian profile and the slope and intercept of the continuum.The best-fit velocity centroids for the H 2 emission lines range between −20 and 2 km s −1 , with no trends as a function of wavelength.A similar range is derived from the fits to the atomic emission lines, which span −25 to 6 km s −1 .The best-fit FWHMs of the emission lines indicate that the features are barely wider than the instrument resolution, making it challenging to detect any kinematic signatures in the 1D line profiles.
Figure 5 shows the emission-weighted velocity maps for the H 2 S(2) and [Ne II] emission lines (see, e.g., Beck et al. 2008;Narang et al. 2024).No velocity gradients are detected in either of the spectral images.We do not identify a red-or blueshifted side of the jet or wind, consistent with expectations for outflowing material moving perpendicularly away from the edge-on disk and therefore the line of sight.We also search for signatures of jet wiggling, which is readily detected over large spatial scales with the Very Large Telescope MUSE and may be caused by perturbations from a misaligned companion or a precessing inner disk (see, e.g., Murphy et al. 2021).We do not resolve any such velocity structure in the knots along the jet axis in Tau 042021, although we note that the difference in surface brightness between the eastern and western disk nebulae has previously been reported as variable and may also be characteristic of a misaligned inner disk (Luhman et al. 2009).Larger mosaics and multiepoch observations will be required to fully characterize jet precession and emission variability in MIRI MRS line images. Observations

Note.
a We allowed the emitting area of the 400 K H 2 component to vary between 0.01 and 200 au, based on the vertically extended spatial distribution in Figure 2. spectral resolution of MIRI at 12 μm (Δv ∼ 120 km s −1 ).For a typical jet speed of ∼150-200 km s −1 (Shang et al. 1998;Narang et al. 2024), the line-of-sight velocity toward a disk with i > 85° (Villenave et al. 2020) would be <20 km s −1 .This is again smaller than the MIRI spectral resolution and consistent with the velocity centroids in Figure 5. Rather, the images of spatially extended wind and jet emission presented here are complementary to high spectral resolution observations that can measure kinematics.

Comparing Mass-loss Rates Derived from MIRI MRS Spectral Images
As reported in Section 3, vertically extended H 2 is observed beyond the scattered-light continuum in the S(1)-S(4) emission lines.This behavior is distinctly different from the CO and H 2 O emission lines, which show emitting regions that are more consistent with the dust distributions in the MIRI MRS wavelength range.Assuming that the observed H 2 S(1) emission is optically thin, we use the integrated line flux to estimate the mass of warm gas contained in the wind (Thi et al. 2001): is the fraction of H 2 molecules in the upper state ( ¢- is the Einstein coefficient describing the probability of radiative decay from J = 3 to J = 1 (Roueff et al. 2019), and Ω is the solid angle spanned by the emitting region reported in Table 2.
The H 2 emission lines in the MIRI MRS spectrum are consistent with two temperature components: T = 400 K, which dominates the S(1)-S(4) emission lines, and T = 1500 K to capture the line fluxes from S(5) to S(8).Assuming the 400 K temperature component in the wind, we measure M M 5.4 10  (Pascucci et al. 2023).However, we note that the measured accretion rate is also 3 orders of magnitude lower than the threshold for strong accretors with HVC [Ne II] emission line components in 1D spectra (Pascucci et al. 2020), which originate in jets like the one imaged with MIRI from Tau 042021.This apparent contradiction likely takes its origin from the fact the the H I line may be strongly impacted by scattering along the line of sight through the edge-on disk, masking the bulk of the emission from the accretion shock itself (see, e.g., Alcalá et al. 2019).Additionally, some of the Huα emission may originate at the base of the jet, and the large methodological uncertainty in deriving a stellar mass for a target in an edge-on disk plays a role as well (see, e.g., Hartmann et al. 2016 for a recent review).However, we note that the line-to-continuum ratios across both sides of the disk are roughly constant, with L/C ∼ 1.3 × 10 −4 in the eastern nebula and L/C ∼ 2.8 × 10 −4 in the western lobe.This effect is not seen in the atomic forbidden line images tracing the more extended jet.It is likely that the linear relationship between H I Huα emission and accretion luminosity does not hold for edgeon disks like Tau 042021, which will require H I and dust scattering models to retrieve the missing line flux.

Distinguishing between Outflow-launching Mechanisms
The semiopening angle of 35°± 5°measured from the H 2 S(2) emission line (Figure 2) is consistent with the range of semiopening angles between ∼20°and 40°for low-velocity winds reported in Pascucci et al. (2023).These values were derived from spectrally resolved [O I] λ6300 emission lines, which can be decomposed into HVCs (v = 50-300 km s −1 ) and LVCs (v < 10 km s −1 ).The largest blueshifts are detected in LVCs from disks with inclinations i ∼ 35°, where the ejected material is moving directly toward the observer along the line of sight (Banzatti et al. 2019).Indeed, MHD wind models predict that mass outflow occurs when the poloidal magnetic field lines reach semiopening angles of θ < 30°relative to the disk rotation axis (Blandford & Payne 1982).Our H 2 observations are consistent with both the wind model predictions and the [O I] LVCs, implying an MHD wind origin for the vertically extended emission.
When spatially resolved, the launching radii of the H 2 wind and the atomic jet can provide further constraints on the outflow-launching mechanism.A compact emitting region is typically associated with an MHD wind (r < 1 au; Fang et al. 2023b), while more extended distributions are characteristic of photoevaporative winds (Bajaj et al. 2024) and jets (Murphy et al. 2021;Whelan et al. 2021).However, the pixel scale of the MIRI channel 3 detector (0 245, or 34 au at d = 140 pc) is large relative to the expected launching radii, and we do not meaningfully distinguish anything inside this minimum inner working angle.
Spitzer-IRS observations of Tau 042021 were used to classify the object as a class II disk with no remaining envelope based on its extinction-free infrared spectral indices (n 5-12 = −0.73,n 12-20 = −0.75;McClure et al. 2010;Furlan et al. 2011).Although 13 CO J = 3 − 2 emission is observed at large scale heights (Simon et al. 2019), it is not coincident with the vertically extended emission near 12 μm (Duchêne et al. 2024), and the submillimeter gas surface density distribution does not show signs of an outflow cavity.The mid-infrared H 2 emission also does not appear similar to the spatially resolved outflow cavities detected in younger protostellar systems (see, e.g., Federman et al. 2023), consistent with the classification of Tau 042021 as a later stage of disk evolution.As reported in Section 4.1, we do not resolve velocity structure in this edge-on disk with the MIRI MRS, but the measured semiopening angle of the spatially extended H 2 emission is most consistent with MHD wind model predictions.This is consistent with models that require MHD winds to reproduce the observed relationship between M acc  and M disk in the disk population (Tabone et al. 2022a;Somigliana et al. 2023).

Summary and Conclusions
We present JWST MIRI MRS observations of gas-phase emission in the edge-on protoplanetary disk system Tau 042021.
The data provide spatially resolved spectral images of gas-phase atomic and molecular emission lines, revealing the following.
1.An X-shaped MHD disk wind with a semiopening angle of 35°± 5°, traced by vertically extended emission from rotational transitions of H 2 (see also Duchêne et al. 2024).2. A collimated, bipolar jet in emission from [Ne II], [Ne III], [Ni II], [Fe II], [Ar II], and [S III] (see, e.g., Duchêne et al. 2014Duchêne et al. , 2024)), with an opening angle of <3°. 3. Radially extended CO and H 2 O consistent with line emission from the inner disk at 1 au, scattered on large dust grains in the disk surface out to 100 au. 4. H I Huα emission in each nebula, consistent with scattered light from the accretion shocks or emission from the base of the jet.
Assuming the H 2 emission is optically thin, the wind contains a gas mass of M M 5.4 10 ´yr −1 , although we note that this is likely an underestimate due to scattering and extinction along the line of sight and uncertainties in the size of the line-emitting region.This work demonstrates the unexpected ability to spatially resolve gasphase emission lines from large protoplanetary disks with the MIRI MRS, which we have shown to be a powerful tool for measuring disk mass loss in winds and jets and exploring inner disk substructure and compositions through scattered light.

Figure 1 .
Figure 1.Left: 1D MIRI MRS spectrum of Tau 042021, extracted from the eastern (top) and western (bottom) reflection nebulae.We detect H 2 emission lines and forbidden emission lines from [Ne II], [Ne III], and [Fe II] on both sides of the disk.H 2 O and CO 2 ice absorptions are also identified, along with PAH emission near 11 μm.Right: zoomed-in spectra of rovibrational CO, rovibrational H 2 O, and rotational H 2 O emission lines from the eastern reflection nebula, with LTE slab models overlaid in cyan (Salyk 2022).The CO model has a temperature and column density of T = 1500 K and N log 17.5 = cm −2 , and the rotational H 2 O emission lines are consistent with T = 850 K and N log 15.0 = .The rovibrational lines are shown with the rotational model scaled down by a factor of 20 in flux.

Figure 2 .
Figure 2. MIRI MRS images of Tau 042021, showing the ALMA 0.89 mm dust continuum (top left; Villenave et al. 2020) and emission lines from H 2 S(2) 12.278 μm (top right), [Ne II] 12.814 μm (bottom left), and H I Humphreys α 12.372 μm (bottom right) detected with the MIRI MRS.White contours in each image show the mid-IR scattered-light continuum at 12.2 μm, and the white dashed lines show the position angle measured from the ALMA continuum image (PA = −16°; Villenave et al. 2020).Images have been scaled using an asinh stretch to maximize the contrast between the background noise and the high-S/N emission from the source.Beige circles in the bottom left corner of each panel represent the average theoretical FWHM of the PSF, calculated from the relationship reported inLaw et al. (2023).The observed spatial distributions are consistent with micron-sized grains in the eastern and western reflection nebulae (12 μm continuum), a disk wind (H 2 S(2)), a bipolar jet ([Ne II]), and scattered light from accretion shocks or from the base of the jet (H I).We note that the position angle of the jet is tilted by ∼2°. 5 relative to the disk.

Figure 3 .
Figure3.MIRI MRS image of 11.3 μm PAH emission, which is spatially colocated with the scattered-light continuum from Tau 042021.While PAH emission is readily detected in disks around more massive stars and in more externally irradiated environments (see, e.g.,Geers et al. 2006Geers et al. , 2007)), it is rare to detect the band within disks around M-type stars.We note that the continuum on the red side of the band is beyond the edge of the channel 2 detector, making it challenging to extract the total surface brightness profile.The image shown here is from a single frame where the feature peaks in emission.

Figure 4 .
Figure 4. MIRI MRS line images of Tau 042021, showing stacked emission lines from the CO v = 1 − 0 (left) and H 2 O v 2 = 1 − 0 transitions, detected with the MIRI MRS.White contours in each image show the mid-IR scattered-light continuum at 5.0 μm.Images have been scaled using an asinh stretch to maximize the contrast between the background noise and the high-S/N emission from the source.Beige circles in the bottom left corner of each panel represent the average theoretical FWHM of the PSF, calculated from the relationship reported in Law et al. (2023).The emission is spatially extended beyond the expected size of the inner disk, r < 1 au, consistent with scattering by micron-sized dust grains in disk surface layers (see, e.g., Duchêne et al. 2024).

Figure 5 .
Figure 5. Emission-weighted velocity maps of H 2 S(2) (left) and [Ne II] emission lines from Tau 042021 (right; see, e.g., Beck et al. 2008; Narang et al. 2024).No velocity structure is detected in either of the maps, as expected for an edge-on disk from which the maximum jet and wind velocity components are not oriented along the line of sight.
typical wind velocities of ∼ 10 km s −1 predicted by models and the measured maximum emission height of 175 au above the disk midplane, this corresponds to a mass flux of M M 6The uncertainties in this measurement are dominated by the size of the emitting region and the gas temperature, which is less constrained without observations of the H 2 S(0) emission line at 28.2 μm.The measured accretion rate, of magnitude lower than the wind mass-loss rate, which contradicts the ratios of M M measured from the 12 CO and [O I] λ6300 LVC emission lines flux of ∼6.5 × 10 −9 M e yr −1 .The Huα emission was used to estimate a mass accretion rate of M I, [Ne III], [Fe II], [Ni II], [Ar II], and [S II].The full list of detected transitions and emission line fluxes is provided in Table

Table 1
Detected Gas-phase Emission Lines

Table 3
Slab Model Retrieval Parameters