An Hα Impression of Lyα Galaxies at z ≃ 6 with Deep JWST/NIRCam Imaging

We present a study of seven spectroscopically confirmed (Lyα-emitting) galaxies at redshift z ≃ 6 using the JWST/NIRCam imaging data. These galaxies, with a wide range of Lyα luminosities, were recently observed in a series of NIRCam broad and medium bands. We constrain the rest-frame UV/optical continua and measure the Hα line emission of the galaxies using the combination of JWST/NIRCam and archival HST/WFC3 infrared photometry. We further estimate the escape fractions of their Lyα photons ( fescLyα ) and the production efficiency of ionizing photons (ξ ion). Among the sample, six out of seven galaxies have Lyα escape fractions of ≲10%, which might be the status for most of the star-forming galaxies at z ≃ 6. One UV-faint Lyα galaxy with an extremely blue UV slope has a large value of fescLyα reaching ≃50%. These galaxies have a broad range of ξ ion over log10 ξ ion,0 (Hz erg−1) ∼ 25.0–26.5. We find that UV-fainter galaxies with bluer UV-continuum slopes likely have higher escape fractions of Lyα photons. We also find that galaxies with higher Lyα line emission tend to produce ionizing photons more efficiently. The most Lyα-luminous galaxy in the sample has a very high ξ ion,0 of log10 ξ ion,0 (Hz erg−1) > 26. Our results support the scenario that Lyα galaxies may have served as an important contributor to cosmic reionization. Blue and bright Lyα galaxies are excellent targets for JWST follow-up spectroscopic observations.

Large ground-based telescopes and Hubble Space Telescope (HST) helped us to find a large number of Lyαemitting galaxies with redshift reaching z 6 − 7, corresponding to the end of EoR.Most of them are Lyα emitters (LAEs) selected by the narrowband (Lyα) technique (e.g., Kashikawa et al. 2006Kashikawa et al. , 2011;;Hu et al. 2010;Shibuya et al. 2018;Taylor et al. 2021).The rest of them are Lyman-break galaxies (LBGs) selected by the dropout technique and identified by Lyα lines (e.g., Steidel et al. 1996;Jones et al. 2012;Inami et al. 2017;Pentericci et al. 2018).The LAEs and LBGs (with Lyα lines) are probably indistinguishable in terms of their intrinsic properties such as age, stellar mass, and star formation rate (SFR; e.g., Dayal & Ferrara 2012;Jiang et al. 2016;de La Vieuville et al. 2020).We thus call both of them as Lyα galaxies in the following text.
To understand how SF galaxies contribute to the ionizing photon budget, their rest-frame optical properties (continua and line emission) are necessary to be constrained.However, this task is difficult to execute before JWST era, especially for z 6 galaxies.For example, due to a lack of near-/mid-infrared (IR) bands, it is challenging to break the degeneracy between prominent nebular emission from young galaxies and strong Balmer breaks from old galaxies (e.g.Schaerer & de Barros 2009;Jiang et al. 2016).On the other hand, even if the galaxies are spectroscopically confirmed (by Lyα line for instance) at z 6, the optical emission lines (mainly [O iii]+Hβ and Hα+[N ii]) simultaneously boost the IRAC1 and IRAC2 channels of Spitzer Space Telescope (e.g.Faisst et al. 2016;Harikane et al. 2018;Stefanon et al. 2021), leaving lines and continua coupled together.Such problems are being well solved in the current JWST era.
For building a large and homogeneous sample of highz galaxies, we have carried out the Magellan M2FS spectroscopic survey to identify LAEs and LBGs at z 6 (Jiang et al. 2017;Ning et al. 2020Ning et al. , 2022;;Fu et al. in preparation).A fraction of them will be covered by the upcoming JWST imaging survey, such as COSMOS-Webb (GO 1727; Kartaltepe et al. 2021;Casey et al. 2022) and PRIMER (GO 1837; Dunlop et al. 2021).The multiple IR bands can reveal their individual properties in detail.Recently, one of our identified LBGs has been covered by parallel JWST/NIRCam imaging of PRIMER MIRI observations.In this work, we compare it with spectroscopically confirmed galaxies at z 6 from from previous literatures (Pentericci et al. 2018) to give a pilot investigation on the Hα properties of luminous Lyα galaxies.
This paper is organized as follows.
In Section 2, we briefly present the sample of Lyα galaxies, JWST/NIRCam imaging observations, data reduction, and photometry.In Section 3, we give the measurement results of the Lyα, ultraviolet (UV), and Hα-related properties of the galaxy sample.We discuss their Lyα escape fractions and ionizing photon production rates in Section 4. We summarize this work in Section 5. Throughout the paper, we use a standard flat cosmology with H 0 = 70 km s −1 Mpc −1 , Ω m = 0.3 and Ω Λ = 0.7.All magnitudes refer to the AB system (Oke 1974).

SAMPLE AND DATA
In this section, we describe our sample of Lyα galaxies at z 6, JWST/NIRCam imaging observations, data reduction, and photometry.We summarize the sample information in Table 1.

Sample of Lyα Galaxies at z 6
The sample includes seven spectroscopically confirmed galaxies at z 6.The first one (SC-1) is confirmed at redshift z = 6.087 with a strong Lyα line by our spectroscopic survey (see the upper panel of Figure 1; Fu et al. in preparation).In this survey, we carried out spectroscopic observations using the fiber-fed, multi-object spectrograph Michigan/Magellan Fiber System (M2FS; Mateo et al. 2012) on the 6.5 m Magellan Clay telescope.The science goal is to build a large and homogeneous sample of high-z galaxies (see Jiang et al. 2017 for an overview of the program), including LAEs at z ≈ 5.7 and 6.6 (Ning et al. 2020(Ning et al. , 2022)), and LBGs at 5.5 < z < 6.8 (Fu et al. in preparation).These high-z galaxies are located in the famous fields including the Subaru XMM-Newton Deep Survey (SXDS), the Extended Chandra Deep Field-South (ECDFS), A370, COSMOS, and SSA22.The total sky area is around 2 deg 2 .SC-1 has an estimated Lyα equivalent width b This value correspond to a 3σ upper limit.
of EW 0 (Lyα) 100 Å.It is thus one of the largest-EW 0 (Lyα) LBGs in a wide area covered by our spectroscopic survey.
The rest six sources (SC-2-7) are from a previous work, Pentericci et al. (2018, hereafter P18).They used VLT/FORS2 to conduct the CANDELSz7 survey, an ESO Large Program, to spectroscopically confirm SF galaxies at z 6 in three HST CANDELS Legacy fields (GOODS-South, UDS, and COSMOS).SC-2 and SC-3 locate in the CANDELS UDS field while SC-4-7 locate in the CANDELS GOODS-South (GOODS-S) field.Among them, SC-2 was detected with no emission line and a continuum discontinuity which is interpreted as a Lyα break (see P18).Due to the limited sensitivity of instruments and depth of observations, we treat this source as a Lyα galaxy with an upper limit of EW 0 (Lyα) from P18. Figure 1 shows redshift distribution of the sample in the lower panel.

Imaging Data
The three sources SC-1-3 in/around the CANDELS UDS field are covered by the JWST Cycle-1 program (GO 1837), Public Release IMaging for Extragalactic Research (PRIMER; Dunlop et al. 2021).The PRIMER survey pays attention to the two key equatorial HST CANDELS Fields (COSMOS and UDS) by delivering 10-band NIRCam+MIRI imaging observations.It owns the parallel NIRCam imaging in eight bands including F090W, F115W, F150W, and F200W in the short wavelength (SW), and F277W, F356W, F444W, and F410M in the long wavelength (LW).Note that the F410M−F444W color is a good indication of Hα emission of the SF galaxies at z ∼ 5 − 6.6.SC-1-3 have been imaged by PRIMER #14 and #22 observations with individual exposure lengths of ∼1.9 hours in the SW bands and ∼0.5 hours in the LW bands.
The other four sources SC-4-7 are covered by the JWST Cycle-1 program (GO 1963), UDF medium-band survey (UDF-MB; Williams et al. 2021).The UDF-MB survey images the (Hubble) Ultra Deep Field (UDF) with a single NIRCam pointing (field of view ∼ 2 × 2 .2×2.2) in a series of medium-bands including F182M, F210M, F430M, F460M, and F480M (NIRISS F430M and F480M in parallel).Total integration time reaches ∼7.8 hours for each F182M, F210M, and F480M and ∼3.9 hours for each F430M and F460M.In Figure 1, we plot the transmission curves of the five JWST/NIRCam LW filters to compare and illustrate the Hα locations of the galaxies in the observed-wavelength frame.Except SC-1 and 2, other sources are not covered by enough JWST SW bands especially SC-3 is located in the gap region of NIRCam SW imaging.We thus utilize the archival HST/WFC3 (Wide Field Camera 3) near-IR imaging data from the CANDELS program (Grogin et al. 2011;Koekemoer et al. 2011;Guo et al. 2013;Galametz et al. 2013).We download the data products provided by the High Level Science Products1 .SC-2 and SC-3 have HST/WFC3 F125W and F160W observations while SC-1 is a little bit outside the CANDELS UDS region.SC-4-7 in the CANDELS GOODS-S region have HST/WFC3 observations in three near-IR bands (F105W, F125W and F160W).

Data Reduction and Photometry
We reduced the NIRCam imaging data with the standard JWST pipeline2 (v1.7.2) up to stage 2 us-ing the reference files "jwst 0999.pmap" for PRIMER and "jwst 1008.pmap" for UDF-MB.Then we use the Grizli3 reduction pipeline to process the output images.Grizli mitigates 1/f noises and mask the "snowball" artifacts from cosmic rays (Rigby et al. 2022).It further converts the world coordinate system (WCS) information in the headers to the SIP format for each exposure so that images can be drizzled and combined with Astrodrizzle4 .For the SW and LW images, the WCS of final mosaics are registered based on the catalogs of DESI Legacy Imaging Surveys Data Release 9 and the pixel scale was resampled to 0.03 with pixfrac = 0.8.We also subtract an additional background on the final mosaics.Figure 2a shows the thumbnail images of the sample in a series of the JWST/NIRCam (and/or HST/WFC3 near-IR) bands.
We run SExtractor (Bertin & Arnouts 1996) to perform photometry in the JWST/NIRCam multi-band images.The aperture has a radius of triple FWHMs of the point-spread function (PSF) in each wavelength band.The aperture correction is calculated from the PSF in each band.We first obtain initial measurements by matching the output catalogs to the targets within a distance tolerance of a FWHM.For each target, we select its brightest band to feed the detection image.Specifically, we adopt the F444W band for three sources covered by the PRIMER survey and the F460W band for the four sources covered by the UDF-MB survey because their Hα lines boost these bands.We then rerun SExtractor in the dual image mode with the detection images.For each measurement image, we also adopt an aperture with a radius of triple PSF FWHMs in this band.Only for the SC-7 source, we use a radius of 1.5 PSF FWHM, in order to minimize the amount of abnormal pixels within the photometric aperture, caused by the fact that SC-7 is located very close to the image edge.Table 1 lists the multi-band photometry results of the galaxy sample.

RESULTS
In this section, we give the measured results of UV and Hα quantities of the galaxy sample.Their UV properties are derived from JWST/NIRCam SW bands and/or HST/WFC3 near-IR bands.We utilize the NIR-Cam medium-bands to constrain their (rest-frame) optical continuum.We then measure their Hα flux by combining the corresponding LW broad-or medium-bands.We further obtain their Hα-related properties including the Lyα escape fraction and the ionizing photon production efficiency.The results are listed in Table 2.

UV Continua
We measure the UV continuum of the galaxies with the commonly used method (e.g., Pentericci et al. 2018;Jiang et al. 2020).As in these works, we assume a power-law form for the UV continuum of each source, i.e.: f λ ∝ λ β .As we work in AB magnitude units, we fit a linear relation m AB ∝ (β + 2) × log(λ) to the SW photometric data, from which we obtain the UV continuum slope β UV and the absolute UV magnitude M UV at the rest-frame wavelength 1500 Å.
For a galaxy at z 6, its observed Lyα line locates in the wavelength range of the Subaru/z and JWST/F090W bands.The corresponding broad-band flux usually differs from the flux level of UV continuum due to the Lyα emission or break (IGM absorption bluewards of Lyα).So in the measurements for UV continua, we abandon the F090W photometric data for SC-1 and SC-2.We then subtract the fit power-law UV continuum from the z -or F090W-band photometry to constrain Lyα flux and EW 0 (Lyα) for SC-1.The IGM continuum absorption blueward of Lyα line is considered in the computation (Madau 1995).For SC-3-7, the CANDELz7 galaxies, we directly adopt the observed Lyα flux given by P18 to compute the Lyα luminosity and EW 0 (Lyα) with our obtained power-law UV continua.For SC-2 which is undetected in Lyα, we use the upper limit of EW 0 (Lyα) given by P18 to compute its Lyα flux.Note that Lyα flux may be slightly underestimated due to the potential Lyα emission from the circumgalactic medium (e.g., Cai et al. 2019;Wu et al. 2020).

Hα Line Emission
We combine the medium-and broad-bands (covering the rest-frame optical wavelength at z 6) to estimate the flux and EW of Hα emission lines.In Figure 2, the (red) color of F410M−F444W and F430M−F460M clearly show the flux excess due to strong Hα lines.In the F444W broad-band which covers the F410M, F430M, F460M, and F480M medium-bands, Hα+[N ii] lines dominate the flux estimation (e.g., Anders & Fritze-v. Alvensleben 2003).We thus ignore other optical lines except Hα+[N ii].We assume that [N ii] contribute line flux at the Hα wavelength due to the small wavelength difference relative to the wavelength range of >4 µm.We also assume that the emission line has a gaussian profile with FWHM = 300 km s −1 (see the black lines in Figure 2b).
SC-4-7 are covered by the UDF-MB survey with five NIRCam medium-bands.For SC-4, 5, and 7, we use the F430M and F480M magnitude (3σ upper limit for SC-7) to constrain the rest-frame optical continuum with a power-law form f ν ∝ ν α because the Hα+[N ii] emission only fall into the F460M band.For SC-6, the Hα+[N ii] emission fall into the F460M and F480M bands.We thus match the continuum plus line model to the three LW medium-band photometric data.Figure 2b illustrates the above procedure in the lower panel.For these four sources, we obtain similar power-law indices with a median value of α ∼ −0.6.SC-1-3 are covered by the PRIMER NIRCam multi-band (7 broad + 1 medium) observations.As no strong nebular lines fall into the F410M band, we use the F410M magnitude to constrain the (rest-frame) optical continuum and the F410M−F444W color to estimate Hα+[N ii] flux.We start with a power-law continuum with an index of α 0 = −0.6 (from the other four sources) to match the F410M flux density.Then we integrate the known continuum plus unknown line emission weighted by the F444W filter transmission curve to match the F444M flux density and compute the Hα+[N ii] flux.We also vary the continuum slope α in a reasonable range of α 0 ± 0.5 to obtain the deviations of the measured line flux which would be included into the errors of the final values.Figure 2b illustrates the above procedure in the upper panel.After estimating the line flux, we assume that Hα accounts for 85% of the Hα+[N ii] flux, which is similar to previous studies (e.g., Rasappu et al. 2016;Faisst et al. 2019;Sun et al. 2022).To be conservative, we also feed the 10% flux into the error of the final Hα flux.With the measured flux of Hα line, we obtain the SFR(Hα) using the canonical Hα-SFR calibration relation (listed in Column 11 of Table 2; Hao et al. 2011;Murphy et al. 2011;Kennicutt & Evans 2012).We further compute the EW 0 (Hα) with the rest-frame optical continuum level.The results are plotted in the second and third rows of Figure 3.Note that the Hα flux and EW may be underestimated because the optical continuum is overestimated due to the existence of some faint optical lines.But such an underestimation is supposed to be included into the enlarged measurement errors.EW 0 (Hα) indicates the specific SFR (sSFR) of galaxies.Our z 6 sample spread a larger range of EW 0 (Hα) than the low-z LAEs (Matthee et al. 2021) and local analogs (Yang et al. 2017).Recently, Sun et al. (2022) serendipitously found a sample of strong Hα/[O iii] emitters in the JWST/NIRCam wide-field slitless spectroscopy (WFSS) data.Our median EW 0 (Hα) is twice higher than theirs because in our sample the 6/7 galaxies emitting Lyα lines are supposed to have higher sSFR while the SF galaxies of Sun et al. (2022) are found based on Hα/[O iii] detections.SC-7 is undetected in Lyα and its EW 0 (Hα) is similar to the lowest one of the sample in Sun et al. (2022).

Escape Fraction of Lyα Photons
We estimate the escape fraction of Lyα photons (f Lyα esc ) for the sample.With the obtained Hα flux, we adopt the canonical ratio L(Lyα)/L(Hα) = 8.7 (e.g., Henry et al. 2015) to calculate the intrinsic Lyα flux and obtain: under the assumption of case-B recombination in T e = 10 4 K (Osterbrock & Ferland 2006).We also apply a dust correction using the reddening law of Calzetti et al. (2000).We can not well constrain the extinction E(B − V ) for nebulae due to a lack of the Balmer decrement (Hα/Hβ) information.Thus, we perform SED fitting using BAGPIPES (Carnall et al. 2018) and obtain A V = 0.8 for SC-2 thanks to the abundance of multiband photometric data for this source.For others, we adopt a lower and modest value of A V = 0.4 as a reasonable assumption because they have higher EW 0 (Lyα) with lower dust content.Note that A V is supposed to be smaller for the six galaxies but the difference is only 0.1 dex for the computed f Lyα esc .The f Lyα esc results are listed in the Column 7 of Table 2.We can see that 6/7 galaxies have f Lyα esc 10% even though they spread over a large range of EW 0 (Lyα) and EW 0 (Hα).This may imply the upper limit of f Lyα esc for most of galaxies at z 6.Note that f LyC esc is supposed to be smaller than the f Lyα esc (e.g., Dijkstra et al. 2016;Izotov et al. 2020).SC-7 has the largest value of f Lyα esc reaching 50% while it is relatively faint in the rest-frame UV (M UV −19) with an extremely blue UV slope (β UV −3).In Section 4.1, we discuss the relevant, potential trends.

Production Efficiency of Ionizing Photons
With the measured flux of UV continua and Hα line, we estimate the Hydrogen ionizing photon production efficiency ξ ion by where Ṅion (s −1 ) is the intrinsic production rate of Hydrogen ionizing photons from stellar populations and L UV ν (erg s −1 Hz −1 ) is the (mono-chromatic) UV continuum luminosity per photon frequency which can be derived from the above measured M UV .ξ ion intrinsically depends on the assumed stellar-population model (e.g., Robertson et al. 2013;Eldridge et al. 2017;Yung et al. 2020b).Ṅion can be computed from Hα emission by Ṅion = L(Hα) in the (T e = 10 4 K) case-B recombination (Kennicutt et al. 1994;Leitherer & Heckman 1995;Madau et al. 1998).We then obtain the production efficiency of ionizing photons which do not escape from the galaxy, ξ ion,0 assuming f LyC esc = 0.The (dust-uncorrected) results are shown in the downmost row of Figure 3.Note that the dust-corrected values are lower by 0.1 dex assuming the canonical stellar/nebular extinction ratio of 0.44 which is obtained from local starbursts (Calzetti et al. 2000).The ξ ion,0 results are listed in the Column 10 of Table 2. Our obtained ξ ion,0 distribute over a broad range of log 10 ξ ion,0 ∼ 25.0−26.5.The median value is consistent with that of a large UV-faint galaxy sample from Prieto-Lyon et al. (2022).This median is also close to that of a Hα-emitter sample from Sun et al. (2022).Among our sample, SC-1 is the most luminous in Lyα and has the highest ξ ion,0 reaching log 10 ξ ion,0 ∼ 26.5 while SC-2 is not detected in Lyα and has the lowest ξ ion,0 .We further give an extensive discussion in next section.

DISCUSSION
Thanks to the excellent capability of JWST (Rigby et al. 2022), we have measured the Hα-related properties of the SF galaxies at z 6.In this section, we discuss the obtained properties of the galaxy sample.We compare our results with previous studies in Figure 3.The Hα-related properties are assigned by rows and Lyα/UV quantities are assigned by columns.The f Lyα esc is assigned again at the rightmost column to compare with the Hα-related properties.

f Lyα esc -UV Correlation
We compare the obtained f Lyα esc in the upmost row and rightmost column of Figure 3.The measured results show that f Lyα esc positively correlates with L(Lyα) and EW 0 (Lyα), and negatively correlates with M UV and β UV (in the upmost row).The positive trends are natural because f Lyα esc is inferred through Lyα flux, which is similar to those found from the low-z samples (e.g., Hayes et al. 2014;Yang et al. 2017;Matthee et al. 2021).Yang et al. (2017) (and Kim et al. 2021) use a statistical sample of local Green Pea galaxies as high-z analogs to reveal that.Matthee et al. (2021) use a LAE sample at z ∼ 2 and also found such a relation.For the negative trends of f Lyα esc changing as UV properties, we obtain a linear relation between f Lyα esc and M UV , shown as a red dotted line in panel 0-0; we also obtain another linear relation between f Lyα esc and β UV , shown as a red dotted line in panel 0-1.The f Lyα esc -M UV relation seems to exist in the current high-z sample, although the sign is weak for the two samples of Lyα galaxies at lower redshifts.Chisholm et al. (2022) found that f LyC esc increases for fainter M UV using a sample of LyC-leaking SF galaxies at z 0.3, which is overall lower than our obtained f Lyα esc -M UV trend.Such difference is consistent with the scenario of f LyC esc f Lyα esc (e.g., de Barros et al. 2016;Dijkstra et al. 2016).
The f Lyα esc -β UV trend may exist at z 6 like at z ∼ 2 (e.g., Snapp-Kolas et al. 2022;Prieto-Lyon et al. 2022).The reason can be simply interpreted as that bluer galaxies have younger stellar population and/or lower dust content.They emit harder ionizing photons suffering lower extinction, which accounts for higher escape fraction of Lyα photons.The z 6 trend is steeper, which is consistent with the current consensus that galaxies at higher redshift are bluer (e.g., Bouwens et al. 2014;Jiang et al. 2020).Chisholm et al. (2022) shows that f LyC esc scales strongly with β UV for the LyCleaker sample at z 0.3, while those with β UV = −2 have an averaged f LyC esc 5% which is smaller than f Lyα esc 10% estimated from our linear fit (panel 0-1 of Figure 3).SC-7 is the bluest one in our sample with f Lyα esc ∼ 50%.This galaxy and the luminous LAEs with very blue UV continua (β UV 3) reported by Jiang et al. (2020) are supposed to be strong LyC leakers to contribute ionizing photons.They are thus excellent targets to carry out JWST IR spectroscopic followup for the ionization lines.
Comparison of the measured Hα-related properties with the Lyα and UV properties.From top to bottom, we show the Lyα escape fraction, Hα luminosity, rest-frame Hα EW, and ionizing photon production efficiency.From left to right, we show the rest-frame UV magnitude, UV slope, Lyα luminosity, and rest-frame Lyα EW.We also compare the Lyα escape fraction with the measured Hα properties in the rightmost column.The red circles indicate the z 6 galaxies measured by this work.The green squares represent the z ∼ 2 LAEs from Matthee et al. (2021).The blue triangles correspond to the local Green Pea galaxies from Yang et al. (2017) and Kim et al. (2021).The red dotted lines indicate the best linear fits to the measured quantities (red circles).
The panel 3-2 (3-3) of Figure 3 shows a potential correlation between the ξ ion,0 and Lyα luminosity (Lyα equivalent width) for the z 6 sample in this work.We obtain a linear relation between ξ ion,0 and L(Lyα): log 10 ξ ion,0 = (0.87 ± 0.19) log 10 L(Lyα) − (11.3 ± 8.2), (6) shown as a red dotted line in panel 3-2.Saldana-Lopez et al. ( 2022) also found a similar trend in a sample of SF galaxies at 3 ≤ z ≤ 5. We notice that such correlations does not always keep for the LAEs at z ∼ 2 from Matthee et al. (2021), especially for those with relative bright Lyα flux.A direct reason is that the z ∼ 2 LAEs in Matthee et al. (2021) generally have higher f Lyα esc , even reaching 1. Matthee et al. (2017b) also obtained several galaxies have f Lyα esc 1 and they explained it using the mechanisms including, for example, Lyα emission is produced by a different mechanism (like cooling radiation; Dijkstra 2014).We look forward to compare their results updated with the additional Hα information.
The ξ ion,0 -L(Lyα) trend can be reasonably expected because harder ionizing radiation of more young stellar populations (higher sSFR shown by the panel 2-2) tend to produce more ionizing photons per nonionizing photons (e.g., Balestra et al. 2013;Mainali et al. 2017) and ionize the ISM more efficiently, even-  tually enhancing Lyα production, transfer, and escape.Note that EW 0 (Lyα) can be inferred to be proportional to ξ ion ×f Lyα esc (e.g., Harikane et al. 2018).This trend implies that galaxies with log 10 L(Lyα) 43.5 (at the bright end of Lyα luminosity function) may have a high ionizing photon production efficiency reaching log 10 ξ ion,0 27.Conversely speaking, SF galaxies with low ξ ion,0 ( 25) may tend to emit negligible Lyα emission.If such a trend exists, the very (Lyα) luminous galaxies could fully ionize their surrounding neutral H i gas with a modest or even small f LyC esc value.This situation is also consistent in which the scenario that luminous Lyα galaxies in the EoR power large ionizing bubbles (e.g., Zheng et al. 2017;Yajima et al. 2018;Hu et al. 2021;Ning et al. 2022).

Implications of High ξ ion
We plot the redshift evolution of ξ ion,0 for the SF galaxies in Figure 4.The red circle symbols indicate our results for the Lyα galaxies at the end of EoR.The figure shows that the ξ ion,0 of SF galaxies increases with redshift.Among the seven Lyα galaxies, three of them (SC-1, 3, and 7) have high production efficiency of ionizing photons of log 10 ξ ion,0 26.This fact reveals the possibility that a portion of galaxies have high values of ξ ion .Maseda et al. (2020) also found an elevated mean ξ ion (log 10 ξ ion 26) in a sample of UV-faint and high-EW LAEs at z ≈ 4 − 5. Finkelstein et al. (2019) presented a reionization model dominated by low-mass and UV-faint galaxies, which would requires such high ξ ion .Our results provide an observation evidence.The high values of ξ ion have not been well explained in galaxy simulations.For example, Wilkins et al. (2016) explored the ξ ion range of EoR galaxies in the BlueTides simulation, in which they used low-metallicity stellar population (with binary systems) models to obtain their highest ξ ion of log 10 ξ ion 26.Yung et al. (2020b) could not also provide such high ξ ion with the semi-analytical models of galaxy formation.Note that their models only predict a rather weak dependence of ξ ion with redshift, which means we have not well understood the physical processes responsible for the redshift evolution shown in Figure 4 (the gray lines).In the meanwhile, a larger sample of fainter galaxies is necessary to be built to overcome the sampling bias at the higher redshift (z > 5).
The median ξ ion,0 of our Lyα galaxy sample is consistent the overall trend (gray lines).Previous results indicate EoR SF galaxies with ξ ion,0 on the overall trend can provide enough ionizing photon budget (e.g., Harikane et al. 2018;Stefanon et al. 2022).Our sample implies that Lyα galaxies may also play a similar role and contribute to the total balance of ionizing photons at z 6.Note that the most luminous Lyα galaxy in the sample, SC-1 (identified by our Magellan M2FS survey), has a very high ξ ion,0 (log 10 ξ ion,0 ∼ 26.5) relative to other SF galaxies even though its Lyα luminosity of log 10 L(Lyα) 43 is only a little larger than the characteristic luminosity of Lyα luminosity function (e.g., Hu et al. 2010;Kashikawa et al. 2011;Zheng et al. 2017;Ning et al. 2022).Its nature need to be revealed based on further spectroscopic observations.Our results support that Lyα galaxies, especially those with intrinsically high EW 0 (Lyα), may significantly contribute the ionizing photons during the EoR.We look forward to comparing our findings with the ionizing photon production efficiency estimated for other Lyα-luminous galaxies (from, e.g., Ning et al. 2020Ning et al. , 2022) ) which will be covered by the JWST observations.

SUMMARY
In this work, we present a pilot study of a spectroscopically confirmed sample of (Lyα emitting) galaxies at redshift z 6 based on the JWST/NIRCam imaging data.The sample includes seven targets: one identified by our Magellan/M2FS spectroscopic survey and six observed by the previous CANDELSz7 survey.All the seven sources we analyze are Lyman break galaxies showing large differences in their Lyα luminosity, ranging from no observed Lyα line up to strong Lyα line of log 10 L(Lyα) 43.3.These objects have been covered by two JWST/NIRCam imaging surveys, PRIMER and UDF-MB, which employ a serious of SW and LW bands.
We have obtained their Lyα-and Hα-related properties by combining the NIRCam broad-and/or mediumbands.Based on the results, we also revealed the potential correlations among the properties of Lyα galaxies at z 6.We summarize our findings as follows: • 6/7 galaxies have Lyα escape fractions of f Lyα esc 10% regardless of their EW 0 (Lyα) and EW 0 (Hα), which might be the status for most of star-forming galaxies at z 6.
• One Lyα galaxy which is relatively faint in the rest-frame UV (M UV −19) with an extremely blue UV slope (β UV −3) has a Lyα escape fraction reaching f Lyα esc 50%.
• Our sample spread over a broad range of the ionizing photon production efficiency over log 10 ξ ion,0 ∼ 25.0 − 26.5 with a median value close to those of the galaxy samples at similar redshifts from other studies.
• Galaxies with more luminous Lyα emission probably have higher production efficiency of ionizing photons (Equation [6]).
• Our identified source (SC-1 analyzed in this study) which is very luminous in Lyα has a very high ionizing photon production efficiency of log 10 ξ ion,0 (Hz erg −1 ) > 26.Its nature merits further investigation.
Our results agree with the scenario in which Lyα galaxies may serve as a significant contributor to cosmic reionization.The bluer and/or more luminous Lyα galaxies are ideal targets for JWST spectroscopic followup observations.We need a larger sample of Lyα galaxies observed by JWST for further analysis on the reionization sources.

ACKNOWLEDGMENTS
We acknowledge support from the National Key R&D Program of China (grant no.2018YFA0404503), the National Science Foundation of China (grant no. 12073014, no. 11721303, and no. 11890693), and the science research grants from the China Manned Space Project with No. CMS-CSST-2021-A05 and No. CMS-CSST-2021-A07.We also thank the anonymous referee for the constructive comments and suggestions that improved this paper.Facilities: HST (WFC3/IR), JWST (NIRCam)

Figure 1 .
Figure 1.Upper panel: The spectrum of the Lyα galaxy (SC-1 with zLyα = 6.087) confirmed by our Magellan M2FS spectroscopic survey.The vertical dashed line marks the observed Lyα wavelength.The shaded region represents ±1σ noise level.Lower panel: The redshift-Hα luminosity distribution of the sample in this work and the transmission curves of the five JWST/NIRCam filters.The highest datapoint corresponds to our confirmed galaxy (SC-1) shown in the upper panel.

Figure 2 .
Figure 2. (a) Thumbnail images of the Lyα galaxies at z 6 in this work.Their ID names are marked at the right end of each row.The size of the images is 2 ×2 (north is up and east to the left).The corresponding band is marked at the top of each thumbnail image (HST/WFC3 bands in blue and JWST/NIRCam bands in red).The medium bands are shown in the bold face.(b) Illustration of measuring Hα+[N ii] line flux of the seven sources (shown in the left figure) including three sources covered by the PRIMER survey (the upper panel gives an example) and four sources covered by the UDF-MB survey (the lower panel gives an example).The boxes with errorbars are photometry in the NIRCam LW bands.They have the same colors as the transmission curves of the five NIRCam filters.The black lines represent the power-law optical continua and gaussian profiles (FWHM = 300 km s −1 ) of Hα lines which boost the corresponding bands.
This work is based on the observations made with the NASA/ESA Hubble Space Telescope and NASA/ESA/CSA James Webb Space Telescope.The HST observations are associated with the Cosmic Assembly Near-IR Deep Extragalactic Legacy Survey (CANDELS) program.JWST data are obtained from the Mikulski Archive for Space Telescopes (MAST) at the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-26555 for HST and NAS 5-03127 for JWST.The JWST observations are associated with programsGO-1837 and GO- 1963.

Table 1 .
Basic information and photometry of the Lyα galaxy sample at z 6 The parenthesis indicates that SC-1 is located close to the CANDELS-UDS imaging region.

Table 2 .
Measured properties of the Lyα galaxy sample at z 6.