SPATIALLY RESOLVED GAS KINEMATICS WITHIN A Lyα NEBULA: EVIDENCE FOR LARGE-SCALE ROTATION

Figures 2 and 3 show the two-dimensional spectra for both the 2009 and 2010 observing runs prior to flux calibration. The Lyα λ1216, C iv λ1550, He ii λ1640, and C iii] λ1909 (as well as [O ii] λ3727 in the 2010 data) are clearly detected in both P.A.s. Continuum emission is detected from the diffuse nebula as well as from several nearby compact sources. The brightest galaxy intersected by the P.A. = 1460 slit (labeled "F") is a foreground source at z ≈ 0.479 object, with unambiguous [O ii], Hβ, and [O iii] emission visible in the LRIS spectroscopy. The second brightest continuum source detected in the P.A. = 1460 data, labeled "A" in Figures 2 and 3, is located at the northwest edge of the diffuse line emission. The one-dimensional spectral extraction at the position of Source A (Figure 4) shows clear Lyα emission and possibly faint He ii emission, likely coming from the nebula rather than from the galaxy itself, but no other strong lines. While there appears to be a hint of emission in the Source A spectrum at the position of Nv at the nebula redshift, the lack of any corresponding emission at C iv or C iii] and the presence of a sky line at exactly the same spectral location leads us to conclude that this is not a real detection of Nv. We therefore have no strong spectral constraints on the redshift of Source A, but due to its proximity to the nebula and the lack of continuum emission shortward of Lyα, it seems likely that Source A is a galaxy associated with PRG1.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In what follows we combine the two years of data. The derived sensitivity functions between the two runs agree reasonably well, and the He ii flux measurements observed from PRG1 generally agree to within the errors for both P.As. Therefore, we combine the data employing a simple variance-weighted mean after binning the 2009 data by two spectrally to match the 2010 data. We focus in this paper on the restframe UV lines; the [O ii] emission will be analyzed in the companion paper on the energetics and physical conditions within PRG1.

Surface brightness profiles are shown for both Lyα and He ii as well as for a stack of the C iv, He ii, and C iii] lines in Figures 5 and 6. Measured above a surface brightness of SB_Lyα ≈ 4.5 × 10⁻¹⁸ erg s⁻¹ cm⁻² arcsec⁻², the nebula spans almost 95 ≈ 80 kpc in diameter along the P.A. = 5244 slit. To compare the relative size of the nebula in each tracer, we measure the diameters containing 50% and 90% of the total line luminosity (Table 3). The surface brightness profiles in Lyα, the non-resonant He ii line, and the C iv+He ii+C iii] composite are all strikingly similar; for example in terms of half-light diameter, D₅₀, the Lyα emission is only slightly more extended than the other rest-frame UV lines, by a factor of ∼1.3. This result is consistent with what we found previously using shallower data (Prescott et al. 2009).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 3. Surface Brightness Profile Sizes

P.A.	Line	D50a	D50	F50b	D90a	D90	F90b
		(arcsec)	(kpc)	(10−17 erg s−1 cm−2)	(arcsec)	(kpc)	(10−17 erg s−1 cm−2)
5244	Lyα	3.05 ± 0.11	25.87 ± 0.94	60.5 ± 2.0	7.14 ± 0.37	60.45 ± 3.14	108.9 ± 3.0
	He ii	2.29 ± 0.16	19.43 ± 1.38	5.1 ± 0.3	6.34 ± 0.84	53.67 ± 7.11	9.1 ± 0.5
	C iv+He ii+C iii]	2.36 ± 0.13	19.96 ± 1.09	11.3 ± 0.6	5.64 ± 0.67	47.73 ± 5.64	20.3 ± 1.0
1460	Lyα	2.29 ± 0.13	19.39 ± 1.08	43.8 ± 2.0	7.00 ± 0.60	59.32 ± 5.10	79.1 ± 3.4
	He ii	1.91 ± 0.17	16.19 ± 1.43	3.9 ± 0.3	4.79 ± 0.59	40.53 ± 5.02	7.0 ± 0.5
	C iv+He ii+C iii]	1.72 ± 0.12	14.54 ± 1.01	7.5 ± 0.5	4.25 ± 0.38	36.01 ± 3.22	13.5 ± 0.8

Notes. ^aDiameter containing 50% or 90% of the total flux measured in a given emission line. ^bTotal flux contained within D₅₀ or D₉₀ in a given emission line.

Download table as:  ASCIITypeset image

Figure 7 shows a comparison between the emission line profiles of Lyα and He ii derived from wide extraction apertures chosen to maximize the total S/N of all four strong lines (72 = 61 kpc and 58 = 49 kpc, respectively, for the P.A. = 5244 and P.A. = 1460 data). Figures 8 and 9 are multi-panel figures showing the same emission line profile comparison but as a function of position along the slit, where the individual panels correspond to subapertures (5 pixels ≈067) spanning the full extent of the nebula. The vertical dashed lines represent the systemic velocity, defined as the centroid of the He ii line at the position where the two slits cross. Our spectral resolution is not sufficient to resolve multiple peaks in the Lyα emission line due to radiative transfer effects, so instead we focus on the centroid offsets of the Lyα line relative to the non-resonant He ii line.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In Figure 10, we plot the velocity offset of Lyα measured in thirteen spatial apertures within the nebula (5 pixels ≈ 067, with a minimum S/N = 3 in both Lyα and He ii) and compare with similar measurements from other galaxy populations. The typical observed offset between the two lines is $\Delta v \equiv v_{\rm Ly\alpha }-v_{\rm He\, \scriptsize{II}} \sim 100\hbox{--}200\, {\rm km s^{-1}}$, depending on position. This low velocity offset is similar to what has been seen in other Lyα nebulae (Yang et al. 2011; McLinden et al. 2013); however, here we are able to probe the kinematics point-by-point within the extended gas to show that the velocity offset is consistently low across the entire ∼80 kpc nebula. The measured velocity offset is less than what is seen in LBGs (Steidel et al. 2010) and more similar to that observed in Lyα-emitting galaxies (McLinden et al. 2011; Hashimoto et al. 2013; Guaita et al. 2013; Song et al. 2014).

Zoom In Zoom Out Reset image size

In Figures 11 and 12, we present the velocity and velocity dispersion profiles for Lyα and He ii. Again, we see very small velocity offsets between Lyα and He ii, typically in the range of ∼100–200 km s⁻¹ to the red. At the same time, the large-scale velocity profile in the P.A. = 5244 slit shows a coherent velocity gradient—∼500 km s⁻¹ over the central 50 kpc of the nebula—while in the P.A. = 1460 slit, the velocity profile is much shallower. While in most spatial apertures the kinematic offset is to the red, in P.A. = 1460 on the side of the nebula closest to Source A there is a hint of a reversal, i.e., Lyα is slightly offset to the blue.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The linewidths measured as a function of position are shown in the lower panels of Figures 11 and 12. Across the P.A. = 5244 slit, the linewidth profile is quite flat, only marginally resolved in Lyα, and unresolved in He ii. On the other hand, in P.A. = 1460 we see a peak in the linewidth profile that is clearly resolved and spatially coincident in both Lyα and He ii. In later sections, we use these linewidths as a measure of the velocity dispersion as a function of position within the nebula. However, given the spatial resolution of our data, we note that these measurements inevitably include a contribution from macroscopic velocity gradients, i.e., of order 80 km s⁻¹ across the 1'' seeing disk for P.A. = 5244.

We expect that Lyα will be optically thick under all but the most extreme situations, yet the consistently small velocity offsets we observe between Lyα and the non-resonant He ii line suggest that Lyα is not being substantially affected by complex radiative transfer in PRG1. This implies either that Lyα is optically thin (which we will demonstrate in this section is not likely the case), or that Lyα is produced in situ over an extended area, thereby largely avoiding the effects resonant scattering would impart.

Before discussing our observations in more detail, it is useful to review how optically thin Lyα can be under plausible astrophysical conditions, particularly under the limiting case of a highly ionized medium.

We start by making the assumption of ionization equilibrium:

$$\begin{eqnarray} &&\alpha _{B} n_{e} n_{{\rm H}\,\scriptsize{II}} = \Gamma n_{{\rm H}\,\scriptsize{I}} \end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray} &&\Gamma \approx Q \sigma _{p} /(4\pi R^{2}), \end{eqnarray} \tag{ 2 }$$

where n_e is the electron number density (cm⁻³), n$_{{\rm H}\,\scriptsize{I}}$ is the H i number density (cm⁻³), n$_{\rm H\,\scriptsize{II}}$ is the H ii number density (cm⁻³), Γ is the photoionization rate (photoionizations s⁻¹), Q is the luminosity of ionizing photons (photons s⁻¹), R is the radius from ionizing source (cm), σ_p is the photoionization cross-section (cm²), and α_B is the Case B recombination coefficient for ${\rm H}\,\scriptsize{I}$.

The Lyα optical depth is:

$$\begin{eqnarray} &&\tau _{{\rm Ly}\alpha } = n_{{\rm H}\,\scriptsize{I}} \sigma _{{{\rm Ly}\alpha }} L, \end{eqnarray} \tag{ 3 }$$

where τ_Lyα is the optical depth at line center, $n_{{\rm H}\,\scriptsize{I}}$ is the H i number density (cm⁻³), σ_Lyα is the Lyα absorption cross-section (cm²), and L is the path through the system (cm).

We make the approximation that H is highly ionized (e.g., in the region around an AGN):

$$\begin{eqnarray} &&n_{\rm H\,\scriptsize{II}{}} \sim n_{{\rm H}} \sim n_{e}. \end{eqnarray} \tag{ 4 }$$

Substituting into Equation (1), we obtain the following relation:

$$\begin{eqnarray} n_{{\rm H}\,\scriptsize{I}} & \approx & \alpha _{B} n_{{\rm H}}^2 / \Gamma\quad \end{eqnarray} \tag{ 5 }$$

$$\begin{eqnarray} & \approx & 4 \pi R^2 \alpha _{B} n_{{\rm H}}^2 / (Q \sigma _{p}). \end{eqnarray} \tag{ 6 }$$

We rewrite this in terms of the ionization parameter U = Q/(4πR²cn_H), where c is the speed of light:

$$\begin{eqnarray} &&n_{{\rm H}\,\scriptsize{I}} \approx \alpha _{B} n_{{\rm H}}/(c \sigma _{p} U). \end{eqnarray} \tag{ 7 }$$

Combining this relation with Equation (3) yields:

$$\begin{eqnarray} &&\tau _{{{\rm Ly}\alpha }} \approx \alpha _{B} n_{{\rm H}} \sigma _{{{\rm Ly}\alpha }} L /(c \sigma _{p} U). \end{eqnarray} \tag{ 8 }$$

We adopt the following values: α_B = 2.59 × 10⁻¹³ cm³ s⁻¹, σ_p = 6.3 × 10⁻¹⁸ cm² at 1 Ryd, σ_Lyα = 5.9 × 10⁻¹⁴ cm² (Storey & Hummer 1995; Verner et al. 1996; Osterbrock 1989), c = 3.00 × 10¹⁰ cm s⁻¹, L = 50 kpc, and n_H = 1.0 cm⁻³.

Taking a range of ionization parameter values (U = 10⁻³–1) results in a Lyα optical depth of:

$$\begin{equation} \tau _{{{\rm Ly}\alpha }}\approx 1.2\times 10^{7}\hbox{--}1.2\times 10^{4} \end{equation} \tag{ 9 }$$

in the limit of high ionization.

In order for τ_Lyα < 1, we would therefore need one of the following to be true:

• 1.  
  U ≳ 10⁴, which is many orders of magnitude higher than what is measured for a typical AGN broad line region (Peterson 1997).
• 2.  
  n_H ≲ 10⁻⁷–10⁻⁴ cm⁻³, which is much less than the (albeit uncertain) existing measurements of n_H for Lyα nebulae (∼1–30 cm s⁻³; e.g., Dey et al. 2005; Prescott et al. 2009).
• 3.  
  L ≲ 5 × 10⁻³–5 pc, i.e., the Lyα is emerging from a very thin skin. This could arise if (1) the Lyα we observe is produced via photoionization within a "blister H ii region" illuminated by an offset source, or if (2) Lyα is produced throughout the cloud but all buried Lyα is efficiently extinguished, i.e., by dust.

Thus, even in the highly ionized limit, τ_Lyα ≫ 1 for most reasonable physical parameters. Given sufficient spectral resolution, we would expect to see evidence for substantial resonant scattering of the Lyα line in the form of a larger spatial extent and/or intrinsically double peaked, complex profiles with kinematic offsets relative to a non-resonant tracer. At lower spectral resolution, this would translate into broadened Lyα emission lines with offsets in the observed line centroid.

How much is Lyα being affected by radiative transfer effects relative to the other emission lines, or similarly, how much is Lyα resonant scattering responsible for the large physical extent of the nebula seen in Lyα? At the spectral resolution of our data, we would expect that any intrinsically complex, multi-peak Lyα profiles to result in the observed Lyα line being broader than a non-resonant tracer. This is consistent with our observation that the Lyα line is broader than He ii in the aperture where both lines are resolved ($\sigma _{{\rm He}\,\scriptsize{II}}\sim 390$ km s⁻¹ and σ_Lyα ∼ 570 km s⁻¹ for Lyα, after correcting for the instrumental resolution; Figure 12). This suggests that, as expected, Lyα is optically thicker and undergoing more complicated radiative transfer than the He ii line.

However, the simple observational fact that a non-resonant line like He ii is seen to be nearly as spatially extended as the Lyα emission is a strong argument that resonant scattering of centrally produced Lyα is not the primary factor responsible for the large spatial extent of the Lyα emission. A similar observation was made in the case of LABd05, a Lyα nebula at z ≈ 2.7 that shows diffuse UV continuum emission comparable in extent to the Lyα (Prescott et al. 2012a). In at least these two systems, Lyα scattering is not the main reason we observe a ∼100 kpc scale Lyα nebula. Instead, the Lyα photons are predominantly being produced in situ within the extended gas. At the same time, polarization data from a different Lyα nebula system suggests a significant contribution from scattered Lyα emission (SSA22-LAB1; Hayes et al. 2011). The prevalence of scattering versus in situ production in Lyα nebulae as a class remains to be quantified, but these few case studies suggest that both mechanisms play a role in producing extended Lyα sources.

Lyα radiative transfer modeling indicates that Lyα photons propagating through outflowing or infalling gas should appear redshifted or blueshifted, respectively, relative to the systemic velocity (e.g., Dijkstra et al. 2006; Verhamme et al. 2006). With this in mind, we can ask whether Lyα shows velocity offsets relative to the non-resonant He ii line in PRG1. In general, we see relatively small offsets between the two lines (typically ∼100–200 km s⁻¹), with Lyα usually shifted to the red (suggestive of an outflow). In the case of a simple outflowing shell model, these offsets would correspond to expansion velocities of ∼50–100 km s⁻¹ (e.g., Verhamme et al. 2006). In addition, a few spatial apertures (located near Source A) show blueshifts of Lyα suggestive of mild infall. As we showed in Section 3.3, these velocity offsets are overall lower than what is seen in typical UV-selected star-forming galaxies, and comparable or perhaps slightly lower than for Lyα-emitting galaxies.

By itself, the low velocity offsets could imply one of several possibilities. The outflow velocities could be intrinsically lower in these systems, or the column density of neutral gas could be low (either globally or due to local patchiness). Alternatively, since the radiative transfer from an extended source of emissivity generically results in observed kinematic offsets that are suppressed relative to the case of a central source (e.g., Verhamme et al. 2006), the low velocity offsets could simply reflect the fact that the Lyα photons in the Lyα nebula are being produced over an extended region, rather than being scattered from a central source. In this scenario, the Lyα profiles in each aperture do not actually encode information about the full velocity structure of the system, but instead reflect only small local velocity offsets between the point of emission of the Lyα photons and the final scattering location. For this reason, the Lyα kinematics closely resemble what is measured using a non-resonant line generated within the same region. Combined with the results of the previous section, it seems clear that both the spatial structure and kinematics of the system are consistent with in situ production of Lyα photons in PRG1. From a purely observational perspective, the fact that Lyα traces the non-resonant He ii line so well suggests that using Lyα alone to do kinematic studies may actually be more reliable in this case than is often assumed.

The deep spectroscopy allows us to probe the kinematics of the diffuse gas using multiple lines with great sensitivity out to large physical scales. A successful physical model for the gas kinematics in PRG1 must be able to explain the following observations: (1) a pronounced monotonic velocity gradient in the P.A. = 5244 data, with a flattening at large radii, and a relatively flat velocity profile in P.A. = 1460, (2) a consistently low velocity dispersion in P.A. = 5244 (σ_Lyα ≲ 300 km s⁻¹, corrected for the instrumental resolution), and (3) a conspicuous resolved peak in the velocity dispersion profiles for both He ii and Lyα in the P.A. = 1460 slit, corresponding to $\sigma _{{\rm He}\,\scriptsize{II}}\sim 390$ km s⁻¹ and σ_Lyα ∼ 570 km s⁻¹.

It is easiest to understand the velocity profile within the system if the nebula represents gas undergoing rotation. To show this, we construct a toy model of a simple thin disk with six parameters: the offset angle between the P.A. = 5244 slit and the major axis of the disk (Θ_off, between −45° and 45°), the disk inclination (i, between 0° and 90°), the maximum velocity of the disk (V_max, between 0 and 600 km s⁻¹), the radius at which the disk reaches V_max (R_max, between 0''and 6''), and the position of the slit crossover relative to the disk center (X_c, Y_c, within ±2'', where a positive offset in both parameters indicates a slit crossover located to the northwest of the disk center). We use a simple Markov Chain Monte Carlo (MCMC) fitting approach and 100,000 iterations to determine the best fit to the velocity profile in both slits, and estimate the posterior distribution for each parameter (Figure 13). While we have shown that Lyα and He ii show very similar behavior overall, there are small differences, particular in the region around Source A. For the purposes of fitting the velocity profiles, therefore, we use the more reliable He ii line and restrict the fitting to only those apertures within the central R < 35 of the nebula.

Zoom In Zoom Out Reset image size

The posterior distributions suggest that the velocity profiles agree reasonably well with a thin disk model. In Table 4, we list the 67% confidence intervals for each parameter, and in Figure 14 we show the range of predicted velocity profiles corresponding to these confidence intervals as well as the predicted profiles for two random draws from the posterior distributions, overplotted on the data. The corresponding disk dynamical masses are of the order of M(< R = 20 kpc) ∼3 × 10¹¹ M_☉. The fact that the major axis of the model disk is roughly aligned with P.A. = 5244 explains the classic "rotation curve" structure seen in the velocity profile, i.e., the monotonic velocity gradient and the flattening at large radii. Similarly, the P.A. = 1460 slit lies roughly along the minor axis, explaining the flatter velocity profile. We note that the surface brightness profiles of PRG1 can be reasonably well fit with an exponential profile, consistent with this picture.

Zoom In Zoom Out Reset image size

Table 4. Toy Model Fit Parameters

Parameter	67% Confidence
	Interval
Θoff	[−30°, −16°]
i	[38°, 77°]
Rmax	[08, 23]
Vmax	[233, 441]
Xc	[−09, −03]
Yc	[−07, −04]

Download table as:  ASCIITypeset image

The velocity dispersion for a simple rotating disk observed at some inclination angle is a combination of the intrinsic velocity dispersion (e.g., due to turbulence) as well as the smearing of the disk rotation within the slit; the latter component typically leads to an observed peak in the velocity dispersion at the disk center. Our simple thin disk model does not allow for a formal prediction of the velocity dispersion, but the spread between the edges of the slit can be taken as an indication of the velocity dispersion that would be measured simply due to rotation smearing. For reasonable models, we find that the spread in velocity sampled by the slit is only of order 200 km s⁻¹ (σ ≲ 100 km s⁻¹), well below the resolution of our data. Thus, if the kinematics were driven solely by rotation in a thin disk, we would not expect to resolve the lines anywhere across the nebula. In addition, the MCMC fitting analysis strongly prefers a center of rotation that is offset with respect to the velocity dispersion peak by about 1''–2'' to the northwest. Together, these facts suggest that the velocity dispersion peak we observe is not the kinematic center of the system, but rather the result of local kinematics, e.g., an outflow from a galaxy or clump within the nebula. It is possible that PRG1 resembles a scaled up version of the "clumpy disks" seen at z ≈ 2 (e.g., Förster Schreiber et al. 2009; Newman et al. 2012), which in some cases show a peak in the velocity dispersion that is offset by several kiloparsecs from the disk center due to the presence of a large star-forming clump driving an outflow. In the case of PRG1, this possibility is supported by the presence of a continuum source near the location of the peak velocity dispersion that is visible in recently acquired Hubble Space Telescope/WFC3 F140W imaging (M. K. M. Prescott et al., in preparation).

Despite its simplicity, the toy model provides a good representation of the velocity profiles observed in PRG1. To estimate the stability of the proposed disk, we plot the ratio of the velocity dispersion (corrected for the instrumental resolution, σ_corr) and the circular velocity (V_c, measured from He ii) as a function of position along the P.A. = 5244 slit (Figure 15); as the P.A. = 5244 slit corresponds roughly to the major axis of the disk and the preferred disk inclination is relatively high, we do not apply any corrections for the disk inclination or azimuthal angle within the disk, i.e., we take V_c ≈ V_obs, the observed velocity. Apertures where the measured linewidth is consistent with the instrumental resolution are shown as upper limits (3σ). Using the one aperture along P.A. = 1460 where both Lyα and He ii are clearly resolved, we compute an approximate "radiative transfer correction" for Lyα, i.e., the factor by which the Lyα linewidth should be scaled down in order to match that of the non-resonant He ii line. Under the crude assumption that this factor can be applied across the entire nebula, this approach provides a means of peering below our instrumental resolution limit. The Lyα measurements are then plotted both with and without this "radiative transfer correction." Following Genzel et al. (2014) and assuming a marginally stable disk, we estimate the approximate Toomre Q parameter, a measure of whether the gas will be unstable to collapse and result in subsequent star formation, where Q_approx ≈ (a/1.4) × (1.0/f_gas) × σ_corr/V_c, with a being a geometric factor with values of [1.0, 1.4, 2.0] for a Keplerian rotation curve, a flat rotation curve, and a solid-body rotation curve, respectively, and with f_gas being the gas mass fraction. Over most of the nebula, we can only report upper limits on the σ_corr/V_c, but in regions where the lines are resolved, we estimate that Q_approx is typically greater than 0.67–1.3, i.e., the critical values below which the gas becomes unstable, even under the assumption of an extremely high gas fraction (f_gas = 1). Thus in most of the apertures where we resolve the Lyα emission line, the nebula appears to be stable against collapse. In a few central apertures as well as at larger radii, however, the gas may be unstable to collapse and subsequent star formation.

Zoom In Zoom Out Reset image size

Deep IFU observations of Lyα nebula such as PRG1 detecting multiple emission lines will be important for shedding further light on the complex kinematics of these systems, but the suggestion from the data presented here is that the gas in PRG1 is undergoing large-scale rotation in a clumpy, turbulent disk.

Recent high resolution numerical simulations of Milky Way mass halos (∼10¹² M_☉ at z = 0, which corresponds to ∼10¹¹ M_☉ at z = 3, assuming the halo growth rate from Neistein et al. 2006) have indicated that newly accreted gas will have high angular momentum, spending 1–2 dynamical times in the outer halo as a "cold flow disk" that extends to many tens of kiloparsecs outside the central galaxy (Stewart et al. 2011, 2013). Although the halo masses of Lyα nebulae are poorly constrained, it is possible that we are seeing a similar phenomenon in Lyα nebulae systems like PRG1 and SSA22-LAB2 (Martin et al. 2014), i.e., the early formation of a large Milky Way mass galaxy or galaxy group. In the case of PRG1, the rotation period implied by our disk modeling (T_rot ≈ 4.9 × 10⁸ yr, assuming V_max ≈ 250 km s⁻¹ at R = 20 kpc) is consistent with the system undergoing a handful of rotations by z ≈ 1.67 (when the age of the universe was ∼3.9 Gyr). Our results also motivate further high resolution theoretical work on the angular momentum of cold gas accretion as a function of halo mass, particularly for the higher mass halos thought to host giant Lyα nebulae (e.g., Prescott et al. 2008; Yang et al. 2009, 2010).

At the same time, our data do not favor the idea that gravitational cooling is the dominant powering mechanism responsible for the Lyα emission in this system. Gravitational infall of lower metallicity gas (i.e., "cold flows") would not be expected to produce such strong He ii emission over such a large spatial extent (Yang et al. 2006; Rosdahl & Blaizot 2012), and the presence of C iv and C iii] indicates the gas is at least somewhat enriched. In addition, cold flow powered Lyα nebulae are predicted to exhibit Lyα emission line profiles with a dominant blue peak, owing to infall (Faucher-Giguère et al. 2010), whereas in PRG1 we find that whether Lyα is observed to be redshifted or blueshifted relative to He ii depends on the position within the nebula, with most locations showing redshifted Lyα emission. In addition, there is still debate as to whether gravitational cooling can provide the Lyα luminosities that are typically observed in Lyα nebulae (e.g., Goerdt et al. 2010; Faucher-Giguère et al. 2010; Rosdahl & Blaizot 2012).

What does seem likely is that the gas reservoir in these regions is being supplied by recent accretion, perhaps coming in with significant angular momentum, but with the gas being illuminated and photoionized by a powerful source of ionizing photons, i.e., highly obscured AGN and star formation that is being fueled by the same accretion event. In PRG1, this scenario would lead to the observed strong, spatially extended Lyα, He ii, and metal line emission, and to the small velocity offset—primarily to the red—that is measured for Lyα.⁸ This scenario would also be consistent with the observation of several Lyα nebulae that appear to be aligned with the filament of galaxies they reside in (Erb et al. 2011). One can imagine that the alignment is due to the preferential flow of material within the filament, perhaps entering a messy, rotating disk as it feeds a growing galaxy or protocluster. Detailed kinematic studies using high spatial resolution IFU observations to look for evidence of rotation or coherent flows within a larger number of Lyα nebulae would be ideal for testing this hypothesis.