JWST/MIRI Detection of Suprathermal OH Rotational Emissions: Probing the Dissociation of the Water by Lyα Photons near the Protostar HOPS 370

Using the MIRI medium-resolution spectrometer on JWST, we have detected pure rotational, suprathermal OH emissions from the vicinity of the intermediate-mass protostar HOPS 370 (OMC2/FIR3). These emissions are observed from shocked knots in a jet/outflow and originate in states of rotational quantum number as high as 46 that possess excitation energies as large as E U /k = 4.65 × 104 K. The relative strengths of the observed OH lines provide a powerful diagnostic of the ultraviolet radiation field in a heavily extinguished region (A V ∼ 10–20) where direct UV observations are impossible. To high precision, the OH line strengths are consistent with a picture in which the suprathermal OH states are populated following the photodissociation of water in its B˜−X band by ultraviolet radiation produced by fast (∼80 km s−1) shocks along the jet. The observed dominance of emission from symmetric ( A′ ) OH states over that from antisymmetric (A″) states provides a distinctive signature of this particular population mechanism. Moreover, the variation of intensity with rotational quantum number suggests specifically that Lyα radiation is responsible for the photodissociation of water, an alternative model with photodissociation by a 104 K blackbody being disfavored at a high level of significance. Using measurements of the Brα flux to estimate the Lyα production rate, we find that ∼4% of the Lyα photons are absorbed by water. Combined with direct measurements of water emissions in the ν 2 = 1 − 0 band, the OH observations promise to provide key constraints on future models for the diffusion of Lyα photons in the vicinity of a shock front.


INTRODUCTION
In the realm of molecular astrophysics, one of the most remarkable results obtained by Spitzer was the detection of highly suprathermal OH rotational emissions.The observed transitions, detected with the Short-Hi module of the Infrared Spectrometer (IRS) toward the Herbig-Haro object HH 211 (Tappe et al. 2008), originate in pure rotational states with rotational quantum numbers N as high as 34 and energies as high as E/k = 2.8 ×10 4 K.They are naturally explained as the "prompt emission" that follows the photodissociation of water via the B −X band (also known as the "second absorption band") by radiation in the 114 -134 nm wavelength range; this spectral region includes the strong Lyα line emitted by fast interstellar shocks.This explanation is supported by both laboratory and theoretical studies of water photodissociation through the B − X band, which indicate that OH states as high as N = 47 can be populated (Harich et al. 2000, van Harrevelt & van Hemert 2000).
Suprathermal OH emissions resulting from the photodissociation of water were subsequently observed in protostellar disks with Spitzer: the protostellar disk of DG Tau, in particular, has been the subject of a detailed analysis by Carr & Najita (2014).
Spitzer could not perform high-spectral resolution observations shortwards of 10µm, the short wavelength cutoff of the Short-Hi module of the IRS, and at shorter wavelengths the Short-Lo module on Spitzer/IRS provided a spectral resolving power, λ/∆λ, of only ∼ 60, which was insufficient to detect suprathermal OH emissions.By contrast, the MIRI MRS spectrometer on JWST provides coverage down to the OH band-head at 9.13 µm (and below), yielding spectra with a spectral resolving power ∼ 3000.This unique capability opens up the possibility of detecting suprathermal OH emission in the 9 -10 µm range, a possibility that has been realized in observations of the Orion Bar reported very recently (Zannese et al. 2023), providing a powerful test of model predictions for the spectrum of the OH prompt emission (e.g.Tabone et al. 2021, hereafter T21).
In this Letter, we discuss JWST/MIRI observations of suprathermal OH emissions in the vicinity of the protostar HOPS 370.HOPS 370, a.k.a.OMC2/FIR3, is an intermediate-mass Class 0/I protostar (Furlan et al. 2016).It is located north of the Orion Nebula in the OMC2 region of the integral shaped filament at an estimated distance of 392 pc (Kounkel et al. 2018, Tobin et al. 2020, hereafter T20).Its central protostellar mass, determined from Keplerian motions, is 2.5 M ⊙ , and its bolometric luminosity is 314 L ⊙ (T20).Extensive observations of HOPS 370 have been carried out with multiple observatories -including Herschel, SOFIA, VLA, ALMA, and now JWST -and together reveal an actively accreting protostar with a bipolar jet/outflow that is orthogonal to a rotating disk of estimated mass 0.05 -0.1 M ⊙ .It powers a large outflow traced in millimeter and far-IR lines, which suggests that it is in a state of rapid accretion (T20; Manoj et al. 2013;González-García et al. 2016;Sato et al. 2023).This outflow consists of both a wide-angle wind and a collimated jet, the latter containing shocks that are also seen in non-thermal radio emission (Osorio et al. 2017).The orientation of the disk, with an estimated radius of 100 au, indicates that this source is observed at a high inclination angle of ∼ 72 • (T20; Federman et al. 2023, and references therein).Luminous shocked knots in the northern outflow lobe are characterized by strong emissions from a variety of molecules and atomic ions detected in our observations, including H 2 , H 2 O, CO, OH, Fe + , and Ne + .
In Section 2, we discuss the MIRI and NIRSpec observations carried out toward HOPS 370 and the methods used to reduce the data.The resultant spectra and spectral line maps are presented in Section 3, with particular emphasis on the suprathermal OH emissions from the shocked knots.
The origin of those emissions is discussed in Section 4, in the context of a model in which water is photodissociated by shock-produced Lyα radiation.A brief summary follows in Section 5.

OBSERVATIONS AND DATA REDUCTION
The observations of HOPS 370 were performed as part of the Cycle 1 medium GO program "Investigating Protostellar Accretion (IPA)," (PID 1802, Megeath et al. 2021), which carried out NIRSpec and MIRI IFU observations toward five protostars spanning five orders of magnitude in luminosity (see Federman et al. 2023).A set of 2 x 2 mosaics was obtained with NIRSpec using the G395M/F290LP disperser-filter combination, which provides coverage of the 2.87 -5.10 µm spectral region at a spectral resolving power λ/∆λ ∼ 1000, and with all channels of the MIRI/MRS to provide complete mid-infrared coverage from 4.9 to 27.9 µm at spectral resolving power that ranged from 1500 to 4000 (Jones et al. 2023).The mosaicking was performed with a 10% overlap and a 4-point dither pattern.The total observing time was about 7.5 hours, including overheads.Further details of the observing strategy have been presented by Narang et al. (2023) and Federman et al. (2023).
For the reduction of NIRSpec IFU data, we utilized JWST pipeline version 1.9.5 and the JWST Calibration References Data System (CRDS) context version jwst 1069.pmap.In our analysis, we identified hot pixels not captured by the JWST outlier detection step by applying a custom out-lier detection algorithm specific to NIRSpec observations.More information on the NIRSpec data reduction and the custom flagging routine can be found in Federman et al. (2023).
The MIRI MRS data reduction utilized JWST pipeline version 1.12.5 along with the JWST CRDS context version jwst 1179.pmap.We used the standard Stage 1 JWST pipeline Detector1Pipeline to reduce the MIRI MRS data starting from uncal data.
In the subsequent Stage 2 (Spec2Pipeline), we performed pixel-by-pixel background subtraction using dedicated background observations.This process effectively removed all identified bad pixels, resulting in background-subtracted cal products.However, we observed extended H 2 S(1) and H 2 S(2) emissions in the dedicated background observations, which led to reduced flux for these lines in the final data.Consequently, we repeated the Spec2Pipeline without background subtraction.In this case, we encountered hot pixels in the detector data, which we removed using the VIP package (Gomez Gonzalez et al., 2017;Christiaens et al., 2023).Furthermore, we performed residual fringe correction during Stage 2 for both scenarios, with and without background subtraction.
In Stage 3 (Spec3Pipeline), the CubeBuildStep was set to band mode, generating separate FITS files for each channel and band.We also generated data cubes without dedicated background subtraction, with the outlier rejection function turned off in these cases.
We measured and applied an astrometric offset calibration to the NIRSpec and MIRI IFU data to improve feature alignment and link the coordinates to the Gaia DR3 standard.The offset measurement process and listed offsets applied with uncertainties are presented in Federman et al. (2023).
Additional data reduction tasks were performed using a suite of Python scripts we developed to (1) extract spectra within a circular region of any specified position and radius; (2) fit and subtract a continuum from the extracted spectra; (3a) fit Gaussian lines to continuum-subtracted spectra obtained from task (2) above; or (3b) fit Gaussian lines with a first-order baseline at each IFU position and for each spectral line we targeted, thereby enabling us to generate spectral line maps.
The second of these tasks (continuum fitting) was accomplished using a procedure that lacked any knowledge of the wavelengths of expected spectral lines.This "zero-knowledge" feature avoids the risk of artificially creating spectral lines where lines are expected.Here, for each spectral channel, we fit a third-order polynomial to the fluxes measured within a 17-channel window centered on that spectral channel (i.e. with 8 spectral channels on either side of the central one).The fit was optimized to achieve the best fit to any 10 of the 17 spectral channels in the window, and the continuum flux value for the central channel was then assigned in accordance with that fit.For spectral regions where lines cover less than 7/17 ∼ 40% of the spectral samples, this procedure yields a reliable separation of the continuum (including instrumental baseline ripples) from the lines.For the third task, Gaussian fitting, we used the Levenberg-Marquardt algorithm; here, the line centroid and width were allowed to vary over a narrow range and the line intensity was allowed to vary freely, as were the continuum level and slope for task (3b).

RESULTS
The IFU data acquired toward HOPS 370 are extraordinarily rich, revealing literally hundreds of spectral lines with a signal-to-noise ratio adequate for mapping.These data have and will be presented and discussed in series of papers, some already published (Federman et al. 2023;Rubinstein et al. 2023;Nazari et al. 2024;Brunken et al. 2024) and some in preparation.Here, we focus on the suprathermal OH lines and a small set of ancillary lines that are directly relevant to their interpretation.

Spectral line maps
In Figure 1, we present maps of several spectral lines: a strong well-isolated water line within the

OH suprathermal emission spectra
In Figure 2, we present the 8.8 -13.4 µm spectra obtained toward the shocked knots within the circular region indicated by the white circle in Figure 1.This aperture has a radius of 0.8 ′′ and is centered at a projected distance of 316 au from the protostar (ALMA position) on the OH emission peak at offset (∆αcosδ, ∆δ) = (+0.1 ′′ , +0.8 ′′ ).The spectral region shown in Figure 2

H 2 rotational diagram and inferred extinction
While the H 2 emissions from HOPS 370 will be discussed in detail in a future publication, their present relevance is simply in providing a valuable extinction estimate.Their usefulness for this -S(8) MIRI maps with 2D-Gaussian kernels of the widths needed to degrade the spatial resolution to a common value for all lines.We then obtained average intensities for each line within the circular aperture indicated by the white circle in Figure 1.
Following Neufeld et al. (2006), for example, we fit the rotational diagram with the sum of two components each in local thermodynamic equilibrium (LTE): a warm component at temperature T w , with an aperture-averaged column density, N w ; and a hot component at temperature T h , with an aperture-averaged column density, N h .These components were allowed to have separate ortho-topara ratios, OPR w and OPR h , yielding six free parameters to describe the rotational state of H 2 .
The line-of-sight extinction was treated as a seventh free parameter that was adjusted, along with the other six, to optimize the fit (red and blue dashed curves).The best-fit values are indicated on Figure 4, and are typical of other protostellar outflows observed with Spitzer (e.g.Neufeld et al. 2006).2To evaluate the sensitivity of our conclusions to our choice of extinction law and aperture size, we have also analysed the H 2 rotational emissions within an aperture of radius 0.4 ′′ instead of 0.8 ′′ and for two additional mid-IR extinction laws that have appeared in the literature.The results are discussed in Appendix A, both as they pertain to the H 2 analysis discussed above and to the OH analysis discussed below.The relative OH lines fluxes favor the KPv5 extinction curve over the alternative extinction laws considered in Appendix A, but the primary conclusions of our study are similar regardless of which mid-IR extinction law or aperture size we adopt.With 24 observed line intensities, the number of degrees of freedom here was N dof = 23.
Our analysis here is closely-related to that of T21.The only difference is that we present the minimum possible photon intensity, I UV , that would account for the absolute intensities of the observed OH emissions if every UV photon were absorbed locally by water.This photon intensity is a factor 4π smaller than the quantity Φ introduced by T21 and referred to there as the column density of H 2 O photodissociated per second.The observational intensities clearly show systematic errors that are not fully captured by the statistical error bars.Assuming (1) that the predicted curve for Lyα photodissociation (blue) represents the true behavior, (2) that the statistical and systematic errors both have Gaussian distributions with dispersions that may be added in quadrature, and (3) that the fractional systematic error has the same r.m.s., ǫ, for all lines, we adjusted ǫ to achieve a reduced χ 2 of unity for the best-fit scaling.
While the blue curve provides an excellent fit to the dependence of the line strengths on N U , one aspect of the T21 predictions is in conflict with the observations.Whereas T21 predict roughly equal populations in the symmetric ( 2 Π 3/2 (e) and 2 Π 1/2 (f )) and antisymmetric states ( 2 Π 1/2 (e) and The N U -dependence of the OH line intensities provides information about the ultraviolet radiation field.The red curve in Figure 5 shows the predictions given by T21 (2021) for a blackbody radiation field at 10 4 K instead of a Lyα radiation field.These tend to overpredict the fluxes for N U < 30 relative to those for N U > 40.For ǫ = 0.105, the minimum reduced χ 2 for this case is χ 2 red = 2.98, implying that the blackbody radiation field is disfavored at the [N dof (χ 2 red −1)] 1/2 σ = 6.7σ significance level.
As we did for the H 2 emissions discussed in Section 3.3, we have also analysed the OH emissions within an aperture of radius 0.4 ′′ instead of 0.8 ′′ and for two additional mid-IR extinction laws that have appeared in the literature.The results are discussed in Appendix A.

Fraction, f w , of Lyα photons absorbed by water
The number of Lyα photons available to photodissociate water may be estimated from the Brα flux observed within the circular aperture centered on the shocked knots.In this analysis, we assume a geometry in which the Brα and OH emissions are generated within the shocked knots and viewed directly rather than as a result of scattering, which is favored due their location in distinct shock knots.After degrading the resolution of the Brα map to have the same HPBW as the OH lines, we obtain a value of 1.4 × 10 −14 erg cm −2 s −1 for the Brα flux.If we apply an extinction correction assuming the value of τ 9.7 obtained in Section 3.3 above, this corresponds to an intrinsic flux of 3.3×10 −14 erg cm −2 s −1 .As discussed in Appendix B, shock models appropriate for this source predict typical Lyα/Brα luminosity ratios of ∼ 900, a factor of several larger than the Case B recombination ratio because collisional excitation preferentially enhances Lyα.This would imply a Lyα flux within the aperture of 3.0 × 10 −11 erg cm −2 s −1 , or equivalently 1.8 photons cm −2 s −1 .This corresponds to an aperture-averaged intensity of 3.8 × 10 10 photons cm −2 s −1 sr −1 , a factor of ∼ 23 times as large as the minimum intensity of UV photons needed to account for the OH line fluxes (Section 4.1 above).Therefore, only a fraction f w = 4.3% of the available Lyα photons would need to be absorbed by water to explain the OH emission.

Interpretation of f w
Lyα photons are unlikely to travel far without being absorbed by dust or water.For Lyα radiation, we obtain a grain absorption cross-section per H nucleus of σ abs (Lyα) = 1.9×10 −21 cm 2 , adopting the wavelength dependence and albedo given by KPv5; here, the overall scaling was chosen to match the average N H /A J ratio of 5.6×10 21 cm −2 mag −1 determined by Vuong et al. 2003 from X-ray absorption observations in several nearby dense clouds3 .The water photodissociation cross-section for Lyα is 1.53×10 −17 cm 2 (Heays et al. 2017, and references therein), and thus the ratio of the water absorption rate to the grain absorption rate for Lyα photons is is the water abundance relative to H nuclei.The corresponding fraction of Lyα photons absorbed by water is f w = R/(1 + R).The estimate of R given above is critically dependent on the (poorly known) properties of grains in the outflow.Indeed, it assumes that grains are present in protostellar outflows -as suggested by Cacciapuoti et al. (2024) and references therein -and moreover that their properties are similar to those in the dense interstellar medium.The water abundance required to explain a given value of f w scales linearly with the adopted value of σ abs (Lyα).
If f w = 0.043 as determined in Section 4.2, and given the grain absorption cross-section assumed above, required water abundance is 5 × 10 −6 , amounting to only ∼ 1% of the gas-phase oxygen abundance4 .This is the average value, x(H 2 O), encountered by the Lyα photons as they suffer repeated scatterings with H atoms and execute a random walk prior to their eventual absorption.In the region of Lyα production, the gas is warm (T > ∼ 6000 K) and/or ionized and the water abundance will be extremely small.But if the photons escape the region where they are produced without being absorbed by dust, then the water abundance could plausibly exceed 10 −4 if all oxygen nuclei were driven into gaseous water and R could exceed unity.In this scenario, the average water abundance is less meaningful, and the quantity f w might primarily reflect the probability that a Lyα photon escapes the warm region where it is originally generated and enters a region where water is abundant.The transfer of Lyα radiation is a complex process (e.g.Neufeld 1990) that can be profoundly affected by velocity shifts associated with shock waves (Neufeld & McKee 1989).We defer a detailed treatment of this process to a future study.

Lower limit on the water abundance from H
The rovibrational water line map shown in the upper left panel of Figure 1 shows just one of several dozen emission lines detected in the H 2 O ν 2 = 1 − 0 band, which collectively have a total equivalent width of ∼ 0.120 µm.Figure 6 shows the 5.8 -7.0 spectral µm region that is dominated by these emission lines.Unless the density is extraordinarily high (n H > ∼ 10 9 cm −3 ), these lines are too strong to be produced by collisional excitation.Colored symbols in Figure 6 show the line positions, with stars denoting transitions of ortho-water and crosses denoting those of para-water.A color code (top left) indicates the minimum energies, E min , of the v = 0 states that must be pumped radiatively to excite each transition.We note here that E min may be smaller than the energy, E L , of the lower state of the observed rovibrational transition, since radiative pumping via a given transition may be followed by radiative decay in a different transition of longer wavelength.Roughly 90% of the water emission emerges in transitions that can be pumped radiatively out of the lowest 9 rotational states of water (i.e.those with J ≤ 2 and E/k < 160 K).This behavior suggests a low rotational temperature within H 2 O v = 0 state, most likely because the states are subthermally populated, and supports the hypothesis of radiative pumping.
Although a full treatment of the H 2 O ν 2 = 1 − 0 emissions is beyond the scope of the present study, we may obtain a lower limit on the mean water abundance, x(H 2 O), by assuming that the observed lines are radiatively pumped by radiation from the protostar and that the observed continuum is radiation from the protostar that has been scattered by dust.The equivalent width of the water lines is then where the sum is taken over all lines in the ν 2 = 1 − 0 band, W H 2 O = 0.120 µm is the total equivalent width (summed over all H 2 O lines), F H 2 O is the wavelength-integrated line flux for a given line, F c is the continuum flux at the line wavelength, σ sca is the grain scattering cross-section per H nucleus, is the number density of water molecules in the lower state, and σ λ (H 2 O) is the H 2 O crosssection for a given rovibrational line (per molecule in the lower state), which has an integral over wavelength λ given by A ul λ 4 /(8πc), where A ul is the spontaneous radiative rate.The equality in equation ( 1) applies only if the pumping lines are optically-thin.Because the KPv5 grain model suggests that σ sca varies only slowly over the band, we may take σ sca as a constant and approximate the sum of A ul λ 4 n l (H 2 O) as A band λ4 n(H 2 O), where A band = 24 s −1 is the total spontaneous radiation rate for the band and λ = 6.3 µm is the average wavelength.Given the grain scattering cross-section per H nucleus at 6.3 µm implied by KPv5, σ sca (6.3) = 1.2 × 10 −23 cm −2 , we then obtain5 x( Our water abundance of 3 × 10 −5 is a lower limit, and -depending on the water linewidths -would likely increase if optical depth effects are included.But even this minimum estimate is a factor 6 larger than the value needed to yield the inferred value of f w (Section 4.3 above).This supports a picture in which most Lyα photons are absorbed by dust in a warm and/or ionized zone very close to where they are created in a fast shock and only a minority escape to the region of significant water abundance that is probed by the water rovibrational emissions.Like the estimate of x(H 2 O) derived in section 4.3 above, this independent estimate of the minimum water abundance, derived from W H 2 O , is also dependent on the grain properties in the outflow; it scales linearly with the value adopted for σ sca (6.3).If grains were depleted in the outflow (while maintaining the ratio of σ sca (6.3) to σ abs (Lyα)), both water abundance estimates would decrease proportionally.Future detailed analyses of the H 2 O ν 2 = 1 −0 spectrum and how it varies spatially will be needed to discriminate between the various mechanisms that release water from ices into the gas-phase: these include thermal desorption, sputtering in shocks, and UV photodesorption.

SUMMARY
We have presented a study of OH in an outflow jet from the HOPS 370 protostar observed with MIRI and NIRSpec as part of the IPA program.
1. We have detected pure rotational, suprathermal OH emissions from the vicinity of the intermediate-mass protostar HOPS 370 (OMC2/FIR3).These emissions are observed from shocked knots in a jet/outflow, and originate in states of rotational quantum number as high as 46 that possess excitation energies as large as E U /k = 4.65 × 10 4 K.Only symmetric A ′ states of OH are observed.
2. The relative OH line strengths are entirely consistent with a picture in which the suprathermal OH states are populated following the photodissociation of water in its B − X band by Lyα radiation produced locally be a fast, ionizing shock.Photodissociation by a blackbody radiation field at 10 4 K is found to provide a significantly worse fit to the relative OH line strengths.
3. Using measurements of the Brα flux to estimate the Lyα production rate in shocked gas near HOPS 370, we find that ∼ 4% of the Lyα photons are absorbed by water.
4. The fraction of Lyα photons absorbed by water implies a mean water abundance (relative to H nuclei) in the range x(H 2 O) ∼ 1 − 5 × 10 −6 , the derived value depending upon the adopted grain properties (Appendix B).This estimate is proportional to the grain absorption cross-section assumed at the Lyα wavelength (121.6 nm), and represents the average abundance within the region where Lyα photons scatter prior to being absorbed by dust or water.
5. Assuming that the H 2 O ν 2 band emissions observed from HOPS 370 are radiatively-pumped and that the continuum is scattered light, we obtain a minimum water abundance in the range x min = 0.7 − 3 × 10 −5 , the derived value depending upon the adopted grain properties (Appendix A).
This minimum value is proportional to the grain scattering cross-section assumed at 6 µm, and would be exceeded if the pumping lines are optically-thick.It is a factor of several larger than x(H 2 O), suggesting that most Lyα photons are absorbed by dust in a warm and/or ionized zone very close to where they are created in a fast shock and that only a minority escape to the region of significant water abundance that emits the water rovibrational emissions we observe.shows very little variation with the adopted aperture size.The second row lists, I U V , the required UV photon intensity if every available UV photon led to a water photodissociation via the B − X band.
In determining I U V , the OH lines were extinction-corrected using the specified extinction curve and the corresponding value of τ 9.7 , and Lyα was assumed to be responsible for water photodissociation.
As discussed in Section 4.1, we assumed equal fractional systematic errors, ǫ, for each OH flux

B. SHOCK MODEL PREDICTIONS FOR Lyα/Brα
We have used publicly-available the MAPPINGS V shock model (Sutherland & Dopita 2017;Sutherland et al. 2018) to estimate the Lyα/Brα luminosity ratio within the shocked region where suprathermal OH emissions were detected.The upper states of these lines may be populated both following recombination of H + and by direct colllisional excitation of neutral hydrogen from the ground state.The Lyα/Brα ratio can significantly exceed the Case B recombination value ∼ 300, particularly for lower velocity shocks where collisional excitation is most important, so the use of shock model predictions is important here.
We ran a grid of models with preshock densities, n 0 , spanning the range 10 −1 to 10 5.5 H nuclei per cm −3 in steps of 0.1 dex; and with shock velocities, v s , spanning the range 30 to 220 km s −1 in steps of 5 km s −1 .The preshock ionization state was determined self-consistently.The preshock magnetic field was taken as 0.5 (n 0 /cm −3 ) 1/2 µG, and undepleted solar abundances were adopted.The collision strengths for [Fe II] fine structure transitions were updated to the values given in the recent study of Tayal & Zatsarinny (2018).
Figure 7 shows contours of the predicted Lyα/Brα luminosity ratio in the plane of v s and log 10 n 0 .
Here, the red, cyan and blue contours show where the Lyα/Brα ratio is predicted to be 1000, 1500,

ν 2
photodissociation of water to produce suprathermal OH emissions; the v = 0 − 0 S(3) line of H 2 at 9.66 µm, one of eight pure rotational lines detected with MIRI/MRS that may be used to estimate the extinction toward the source; the [Fe II] 5.34 µm line, a transition recently shown byNarang et al. (2023) to be an excellent tracer of collimated jets in another IPA target source, IRAS 16253-2429; covers 24 securely-detected lines of OH, originating in states with N U ranging from 23 to 46, along with 5 fine structure lines of [Ni II], [Co II], [Cl I] and [Ne II], and two pure rotational lines of H 2 , S(2) and S(3).

Figure 1 .
Figure 1.Spectral line maps obtained toward HOPS 370, shown with a logarithmic stretch.RA and Dec offsets are given in arcsec relative to the ALMA source position (green star).The white circle demarks a 0.8 ′′ radius region centered on the shocked knot.The maps are masked near a bright continuum source in the south where the line fits are unreliable.The red circles shown the beam size (HPBW).

Figure 2 .
Figure 2. 8.8 -13.4 µm spectra obtained toward the shocked knots.From top to bottom: Band 2B spectrum with continuum fit in blue; continuum-subtracted Band 2B, 2C, and 3A spectra.Red numbers above the OH lines indicate the value of N U .

Figure 3 .
Figure 3. Spectra of suprathermal OH lines observed toward the shocked knots.Yellow: observed spectrum.Line positions are marked with vertical lines for each component of the OH quartet, with the same color-coding as in Figure 2. Black histogram: Gaussian fit to the A ′ components (see text).
purpose arises because the S(3) line lies close to a local maximum in the extinction curve -associated with the silicate absorption feature -and therefore provides excellent leverage on the line-of-sight extinction.Using the intensities of the S(1) through S(8) pure rotational lines of H 2 , measured with MIRI/MRS, we constructed the rotational diagram shown in Figure 4. Here, we convolved the S(2)

Figure 4 .
Figure 4. H 2 rotational diagram obtained toward the shocked knots.Black points: no reddening correction.Blue and red points: reddening correction applied.Blue and red dashed lines: best fits to the rotational diagram for ortho and para-H 2 .
photodissociations in the B − X band.These calculations, which rest upon theoretical calculations of the photodissociation dynamics(van Harrevelt & van Hemert 2000) and on experimental measurements(Harich et al. 2000), were presented by T21 for four different radiation fields.Those expected following water photodissociation by Lyα radiation are shown by the blue curve.There is only one free parameter in this comparison: an overall vertical scaling that is proportional to the photodissociation rate within the beam.If every available UV photon led to a water photodissociation via the B − X band, the required UV photon intensity would be I UV = 1.67 × 10 9 photons cm −2 s −1 sr −1 .

Figure 5 .
Figure 5. OH photon intensity as a function of N U .Red points: observed intensities.Also shown are the Tabone et al. (2021) predictions for H 2 O photodissociation by Lyα radiation (blue) and by a 10 4 K blackbody (red).

2
Π 3/2 (f )) of OH, the observations indicate that the symmetric A ′ states are strongly favored; indeed, the antisymmetric A ′′ states are not detected and are at least a factor ∼ 10 less populated than the A ′ states.Regardless of the relative rates at which the symmetric and antisymmetric states are populated, the predictions presented in Appendix D of T21 are expected to apply to the total emission in all four N U → N U − 1 transitions (T21).This behavior, also noted in the recent paper ofZannese et al. (2023), is in fact entirely consistent with a recent theoretical study of the photodissociation process byZhou et al. (2015), which indicates a population ratio A ′ /A ′′ ∼ 40 at the Lyα photon energy (their Figure7).The astrophysical data thus provide a clear confirmation of the molecular physics.A less-pronounced difference (∼ factor 2) between the line fluxes for the A ′ and A ′′ transitions had previously been measured byCarr & Najita (2014) in Spitzer observations of the protostellar disk in DG Tau.These authors discussed the effect in detail, with reference to two possible origins for OH: photodissociation of H 2 O in the B − X band, and chemical pumping following formation via reaction of O( 1 D) with H 2 .The larger difference observed in HOPS 370 may indicate that chemical pumping is relatively less important in this source, at least for the N U ≥ 20 transitions discussed here.

Figure 6 .
Figure 6.Average 5.8 -7.0 µm spectrum obtained toward the shocked knots, with rovibrational transitions of ortho-and para-water marked with stars and crosses.
measurement and adjusted ǫ to yield a reduced χ 2 of unity when comparing case KP1 and KP2 fluxes with the predictions for Lyα photodissociation.The third row lists the reduced χ 2 obtained for each of the six cases.The fourth row lists the corresponding values obtained for a 10 4 K blackbody radiation field instead of Lyα.The fifth and sixth rows indicate the significance with which each case is disfavored relative to KP1 or KP2 with photodissociation by Lyα.The values plotted here indicate (1) that the KP extinction law yields a significantly better fit to the data than either WD or MM; (2) for any extinction curve, assuming photodissociation by Lyα radiation yields a significantly better fit to the data than does assuming photodissociation by a 10 4 blackbody.
and 2000, and black contours show intermediate values spaced by 100.The [Ne III] 15.6 µm to [Ne II] 12.8 µm flux ratio is an excellent tracer of shock velocity.The observed, extinction-corrected value of 0.026 is obtained for shock parameters lying along the locus marked with the green solid curve.The green band indicates the region where the predicted value lies within a factor 1.5 of that observed, with the dotted/dashed boundaries applying to larger/smaller line ratios.As a probe of the preshock density, we have considered the [Fe II] 17.9 µm to [Fe II] 5.3 µm flux ratio, which has an observed extinction-corrected value of 4.17.The magenta curves and magenta band represent analogous results for the [Fe II] line ratio.The constraint on density is less tight, as indicated by the width of the magenta band, and must be considered less reliable because recent independent estimates of the collision strengths show significant differences.Nevertheless, the intersection of the green and magenta solid lines suggest that Lyα/Brα ratio of 900 is consistent with these diagnostic line ratios.For any preshock density in the range 10 2 to 10 5 cm −3 , the flux [Ne III] 15.6 µm to [Ne II] 12.8 µm flux ratio alone suggests a Lyα/Brα ratio in range 700 -1050.

Figure 7 .
Figure 7. Contours of the predicted Lyα/Brα luminosity ratio in the plane of v s and log 10 n 0 .Red, cyan and blue contours: Lyα/Brα = 1000, 1500, and 2000.Green band: region where the predicted [Ne III] 15.6 µm to [Ne II] 12.8 µm flux ratio lies within a factor 1.5 of that observed.Magenta band: region where the predicted [Fe II] 17.94 µm to [Fe II] 5.34 µm flux ratio lies within a factor 1.5 of that observed.
6. Suprathermal OH emissions promise to help elucidate the processes whereby Lyman α radiation first escapes from fast shocks and then enters nearby water-rich surroundings where water has been released from grain mantles by radiative heating or slower non-dissociative shocks, or produced in the gas-phase by neutral-neutral reactions that are slow at low temperatures but rapid at the elevated temperatures attained behind shock fronts.Detailed models, beyond the scope of this Letter, will be needed to understand the transfer of Lyα radiation and to fully model the rovibrational water emissions observed from HOPS 370.APPENDIX A. DEPENDENCE ON ADOPTED EXTINCTION CURVE AND APERTURE SIZE We have evaluated the sensitivity of our conclusions to our choice of extinction law and aperture size.Results are presented in Table 1 for six cases.We considered three different extinction curvesdenoted KP (KPv5), WD (Weingartner and Draine, 2001, with the modifications described in Draine 2003); and MM (McClure 2009) -and two different aperture radii (0.8 ′′ and 0.4 ′′ , denoted by 1 and2).The standard model, adopted in the main text, is KP1 (KPv5 extinction curve with the 0.8 ′′ radius aperture).Different rows in Table1show the values obtained for key parameters for all six cases.The first row below the horizontal line lists the optical depth at 9.7 µm, τ 9.7 , derived from our fit to the H 2 lines.It ranges from 1.53 to 3.12, with the MM extinction law (which is significantly less dominated by the silicate peak) requiring the largest τ 9.7 and the WD extinction law requiring the smallest, but