Multi-wavelength observations of the obscuring wind in the radio-quiet quasar MR 2251-178

The Neil Gehrels Swift Observatory (Gehrels et al., 2004) provides the most efficient way to track the long-term variation of the X-ray and UV flux of AGN. The X-ray flux was provided by the X-ray telescope (XRT, Burrows et al., 2005) operating in imaging photon counting (PC) mode. The soft ($0.3-1.5$ keV) and hard ($1.5-10$ keV) band flux were obtained with the online XRT pipeline (Evans et al., 2009). The UV flux was covered by the UltraViolet and Optical Telescope (UVOT, Roming et al., 2005). The UVW2 (centered around 1928 Å) flux, which historically has shown the strongest UV variability, was used here. The uvotsource tool was used to perform aperture photometry.

The obscurer can absorb a large fraction of the soft X-ray flux, thus, leading to an elevated X-ray hardness ratio. Simultaneous UV and X-ray data enable us to distinguish variability caused by an obscurer from that of the AGN continuum (e.g., Mehdipour et al., 2016b, 2017, 2022b).

For MR 2251-178, archival Swift observations date back to 2014-01-09 (Figure 1). Since the summer of 2020, the Swift X-ray hardness ratio has been gradually increasing (Figure 2). Therefore, we requested multi-wavelength spectroscopic observations of MR 2251-178 with facilities on the ground and in space from September to December 2020, including a coordinated campaign with XMM-Newton, HST/COS, and NuSTAR on 2020-12-16, to confirm the obscuration event and to study the nature of the obscuring material. The Swift X-ray hardness ratio of MR 2251-178 remained at a relatively high level up to 2021-01-05. The target was out of visibility between January and April 2021. In May 2021, the hardness ratios returned to the historical average value.

XMM-Newton (Jansen et al., 2001) observations provide high-quality X-ray spectra in the energy band of $0.3-10$ keV. The Reflection Grating Spectrometer (RGS, den Herder et al., 2001) provides high-resolution soft X-ray spectra below $\sim$2 keV, which can reveal discrete emission and absorption features. The positive-negative junction (pn) CCD (charge-coupled device) camera of the European Photon Imaging Camera (EPIC) instrument can cover hard X-ray spectra in the $2-10$ keV energy band. Adding EPIC/MOS data does not significantly increase the statistics and will complicate the analysis due to cross-calibration among different instruments. We used the XMM-Newton Science Analysis Software (SAS) v19.0 for data reduction following standard procedures. We also include an XMM-Newton observation of MR 2251-178 obtained on 2002-05-18 (observation ID: 0012940101) for comparison purposes. Both $7-38$ Å RGS data and $1.55-10$ keV EPIC/pn data were used for spectral analysis (Sect. 3). To correct for the cross-calibration between RGS and EPIC-pn spectra, the latter were re-scaled by a factor of 0.950 (2002) and 0.966 (2020) to match the flux level in the $7-8$ Å overlapping wavelength range.

NuSTAR (Harrison et al., 2013) complements X-ray spectra obtained with XMM-Newton in the energy band up to $79$ keV. Our NuSTAR data were reduced with the NuSTAR Data Analysis Software (NUSTARDAS) and CALDB version 20201101. The FPMA and FPMB source spectra were extracted from a circular region centered on the point source with a radius of 80 arcsec. The background spectra of equivalent extraction radius were extracted from a nearby source-free region. The $5-78$ keV NuSTAR data were used for spectral analysis (Sect. 3). To correct for the cross-calibration among spectra collected by XMM-Newton and NuSTAR, the latter were re-scaled (with respect to RGS) by a factor of 0.962 (FPMA) and 1.045 (FMPB) to match the flux level in the $5-10$ keV overlapping energy range.

The Cosmic Origin Spectrograph aboard the Hubble Space Telescope (HST/COS, Green et al., 2012) provides high-quality FUV spectra in the wavelength range $1070-1800$ Å at a resolving power of $R\sim$15000. We use both the archival observations in 2011 and the new single-orbit observation in 2020. The archival observations used grating G130M with central wavelength settings of 1291 Å, 1300 Å, 1309 Å, and 1318 Å, and G160M with central wavelengths settings of 1589 Å, 1600 Å, 1611 Å, and 1623 Å. In 2020, we used G130M at the 1222 Å central wavelength setting and all four FPPOS locations along with G160M at central wavelengths settings of 1533 Å and 1589 Å. The multiple central wavelengths and FPPOS settings permit our spectra to span the gaps between detector segments and enable us to eliminate detector artifacts and improve the flat-field properties by sampling each wavelength at different detector locations. We retrieved calibrated data from the Mikulski Archive for Space Telescopes (MAST) and reprocessed with COS pipeline version 3.3.10, which incorporates all improvements related to wavelength scales, error propagation, and time-dependent sensitivity. Measurement of the centroids of interstellar absorption features revealed no need for any zero-point corrections to the wavelength scale, which is accurate to 5 $\rm km~{}s^{-1}$ (Dashtamirova & Fischer, 2020).

To extend the archival 2011 spectrum to shorter wavelengths, we used FUSE observations from 2001 with optimal reprocessing as described by Kriss (2006), and scaled the flux of the FUSE spectrum to match that of COS in the overlapping wavelength range of $1135-1180$ Å. An archival HST Faint Object Spectrograph (FOS) observation from 1996 was also used for comparison purposes.

We obtained NIR spectroscopy on three occasions. To our knowledge, these data are the first NIR spectra of MR 2251$-$178. On 2020-09-06, we used the SpeX instrument (Rayner et al., 2003) on NASA’s Infrared Telescope Facility (IRTF) equipped with the short cross-dispersed mode (SXD, $0.7-2.55~{}\mu$m) and a $0.3^{\prime\prime}\times 15^{\prime\prime}$ slit, which we oriented at the parallactic angle. This setup gives an average spectral resolving power of $R=2000$. The on-source exposure time was $60\times 120$ s and the weather was photometric with a seeing of $0.46$ arcsec, resulting in a high-quality spectrum with an average continuum $S/N\sim 60$. Before the science target, we observed the nearby (in position and airmass) A0 V star HD 218639 and used this standard star to correct our science spectrum for telluric absorption and for flux calibration. We reduced the data using Spextool (version 4.1), an Interactive Data Language (IDL)-based software package developed for SpeX (Cushing et al., 2004) which carries out all the procedures necessary to produce fully reduced spectra.

On 2020-09-30, we obtained a similar cross-dispersed NIR spectrum with the Gemini Near-Infrared Spectrograph (GNIRS; Elias et al., 2006) at Gemini North in queue mode (Program ID: GN-2020B-FT-111). We chose a slit of $0.3^{\prime\prime}\times 7^{\prime\prime}$, which we oriented at the parallactic angle, and obtained an on-source exposure time of $16\times 120$ s. This setup resulted in a spectrum with an average resolution of $R=1400$ and a continuum $S/N\sim 100$. We reduced the data using the Gemini/IRAF package (version 1.13) with GNIRS specific tools (Cooke & Rodgers, 2005). We again selected HD 218639 as our standard star for telluric correction and flux calibration.

Then, one week after the coordinated XMM-Newton, HST/COS, and NuSTAR observations, we obtained on 2020-12-23 a final cross-dispersed NIR spectrum, with TripleSpec (Wilson et al., 2004) at the Palomar 5 m. The entrance slit for this instrument is fixed at $1^{\prime\prime}\times 30^{\prime\prime}$, which gives a spectral resolving power of $R=2700$. The weather was good and the on-source exposure time of $32\times 120$ s resulted in a spectrum with a continuum $S/N\sim 15$. The somewhat reduced quality of this spectrum relative to those from the IRTF and Gemini is due to a high airmass ($\sec z=1.782$) and a higher spectral resolution.

We obtained optical spectroscopy on two occasions. On 2020-10-18, we used the Low Resolution Imaging Spectrometer (LRIS; Oke et al., 1995) mounted on the Keck 10 m telescope equipped with the 600/4000 and 400/8500 gratings for the blue and red arms, respectively, and the $1.5^{\prime\prime}$ slit. This setup gives a relatively large spectral coverage of $\sim 3100-10300$ Å, with a very small spectral gap of $\sim 40$ Å between the two arms. The average spectral resolving power is $R=750$. The slit was rotated to the parallactic angle, but note that LRIS has an atmospheric dispersion corrector. The on-source exposure time was $300$ s at an average airmass of $\sec z=1.349$, and resulted in an average continuum $S/N\sim 70$. The data were reduced using standard longslit routines from the IRAF software package.

Five days before the coordinated multi-wavelength observations, we obtained on 2020-12-11 an optical spectrum at the Palomar 5 m with the Double Spectrograph (DBSP) instrument(Oke & Gunn, 1982). Similar to LRIS, a dichroic splits the light into separate blue and red channels and we chose for these the 600/4000 and 316/7500 gratings, respectively. We observed through a $2^{\prime\prime}$ slit oriented at a parallactic angle for a total exposure time of $300$ s. This setup resulted in a spectrum with an average resolving power of $R=750$ and a continuum $S/N\sim 25$. The night was not photometric, hence, the entire DBSP spectrum needs to be increased by a factor of 1.7 to match the O [iii] $\lambda 5007$ line peak of Keck/LRIS.

The AGN continuum above the Lyman limit can be described as a disk blackbody model. We use the dbb in SPEX, which is a geometrically thin but optically thick Shakura-Sunyaev accretion disk (Shakura & Sunyaev, 1973). Its normalization and temperature are left free during the fitting. The Milky Way reddening along our line-of-sight is taken into account with frozen parameters $E(B-V)=0.039$ (Schlegel et al., 1998) and $R_{V}=3.1$, as used by Nardini et al. (2014).

We use line-free zones in both the HST/COS spectrum ($\sim 1135-1205$, $1356-1364$, and $1428-1442$ Å) obtained on 2020-12-16 and the Keck/LRIS spectrum ($\sim 6000-6050$, $6530-6580$, $7350-7400$, and $7950-8050$ Å) obtained on 2020-10-18 to constrain dbb parameters. To account for the host galaxy emission in the Keck spectrum, a template starlight emission from the bulge (Kinney et al., 1996) was included as a file model with its normalization free to vary (Mehdipour et al., 2015). The modeled luminosity of the host galaxy emission is $1.47\times 10^{43}~{}{\rm erg~{}s^{-1}}$ in the $1000-10000$ Å wavelength range. The dusty torus (NIR excess in Figure 4), Balmer continuum and blended Fe ii emission (the blue bump near UV) are not included in our modeling.

To model the AGN continuum below the Lyman limit, we use a Comptonized disk component (comt, Titarchuk, 1994), a power-law component, and a neutral reflection component (refl, Magdziarz & Zdziarski, 1995; Życki et al., 1999). The warm Comptonization (comt) is one of the possible interpretations of the soft X-ray excess in AGN (Crummy et al., 2006; Dauser et al., 2010; Done et al., 2012; Petrucci et al., 2013). As shown in our previous works (e.g., Mehdipour et al., 2015, 2017, 2021), it suits our purpose of building a broadband SED for photoionization modeling. The power-law component was cut-off exponentially below the Lyman limit and above 100 keV (Orr et al., 2001). We refer readers to A. Gonzalez et al. (in prep) for a detailed study on the X-ray continuum and its variability over the past two decades. The Galactic absorption in the X-ray band was modeled with a hot model in SPEX. The line-of-sight hydrogen (H i and ${\rm H_{2}}$) column density was fixed to $2.71\times 10^{20}~{}{\rm cm^{-2}}$ (Willingale et al., 2013).

The photoionization model pion (Miller et al., 2015; Mehdipour et al., 2016a; Mao et al., 2017) in SPEX was used for both the warm absorber and emitter (e.g., Mao et al., 2018, 2019). Similar to Reeves et al. (2013), three pion absorption components are required with distinct parameters (hydrogen column density, ionization parameter, and kinematics) for the warm absorber. For the warm emitter, two pion emission components are required.

As shown in Figure 3, while the flux in the optical, UV, and hard X-ray bands in 2020 are comparable to those in 2002, the soft X-ray flux was significantly lower due to the presence of the obscurer. The obscurer was modeled as an additional pion absorption component for the 2020 data set.

After taking into account all the obscuration, absorption, emission, and extinction effects, we derive the best-fit parameters of the broadband SED parameters for both 2002 and 2020 (Table 2). We used relatively simple spectral model components to construct the SED, which is sufficient for the purpose of photoionization modeling. In Figure 4, we illustrate the derived broadband 2020 SED model from NIR to X-ray.

The dbb parameters in Table 2 are obtained with NIR to UV data as described in Section 3.1. They are kept frozen for the X-ray analysis (Section 3.2). The dbb normalization equals to $R_{\rm in}^{2}~{}\cos i$, where $R_{\rm in}$ is the radius at the inner edge of the disk, and $i$ is the inclination angle of the disk. Here we assume $i$ equals to the reflection angle $24^{\circ}$ measured by Nardini et al. (2014). Following Mehdipour et al. (2021), with our best-fit dbb parameters (Table 2), we obtain $R_{\rm in}=0.10$ ld (or $\sim 4.4~{}R_{S}$) and a mass accretion rate of $\sim 2.5~{}M_{\odot}~{}{\rm yr^{-1}}$ for MR 2251-178.

From our broadband SED modeling, the bolometric luminosity of MR 2251-178 is $\sim 1.71\times 10^{45}~{}{\rm erg~{}s^{-1}}$ in 2002 and $\sim 1.51\times 10^{45}~{}{\rm erg~{}s^{-1}}$ in 2020. With the black hole mass of $(2.0\pm 0.5)\times 10^{8}~{}M_{\odot}$ (Lira et al., 2011), which is consistent with $\sim 2.4\times 10^{8}~{}M_{\odot}$ (Dunn et al., 2008), the Eddington ratio was $\sim 6.8$ % in 2002 and $\sim 6.0$ % in 2020. In 2011, using the XMM-Newton data (OM, RGS, and EPIC/pn), Nardini et al. (2014) derived a bolometric luminosity of $\sim(5-7)\times 10^{45}~{}{\rm erg~{}s^{-1}}$, which corresponds to an Eddington ratio of $\sim 20-28$ %¹¹1Nardini et al. (2014) adopted a black hole mass of $2.4\times 10^{8}~{}M_{\odot}$ (Dunn et al., 2008), which led them to report an Eddington ratio of $\sim 15-25$ %. .

In Figure 5, we show the best fit to the observed X-ray data, as well as the transmission of the Galactic absorption, warm absorber (X-ray), and obscurer (2020 only). The best-fit $C$-statistics (Kaastra, 2017) are given in Table 2, which includes the total $C$-statistics and statistics of individual instruments.

The presence of the obscurer can significantly lower the soft X-ray flux so that the warm absorber features are less prominent. Therefore, the warm absorber parameters are better constrained with the X-ray spectrum obtained in 2002, where no obscurer was present along our line-of-sight. The best-fit parameters of the warm absorber are given in Table 3. These parameters are consistent within the $1\sigma$ uncertainty of those reported by Reeves et al. (2013).

The hydrogen column density ($N_{\rm H}$), microscopic turbulence velocity ($v_{\rm mic}$), and outflow velocity ($v_{\rm out}$) of the warm absorber derived from the 2002 X-ray spectrum are kept frozen when analyzing the 2020 X-ray spectrum. The warm absorber is expected to be less ionized in 2020 because photons from the central engine are screened by the obscurer before reaching the warm absorber. Since the warm absorber features are not detectable with the diminished soft X-ray flux, following our previous analysis on NGC 5548 (Kaastra et al., 2014) and NGC 3783 (Mehdipour et al., 2017; Mao et al., 2019), we simply assume that the product $n_{e}~{}r^{2}$ of the warm absorber is constant, where $n_{e}$ is the number density and $r$ the distance of the warm absorber to the central engine. For the warm absorber in 2020, the ionization parameter (Tarter et al., 1969; Krolik et al., 1981)

is then calculated with the $1-1000$ Rydberg ionizing luminosity. In 2020 (Table 2), $\log\xi=$1.96, 0.79, and $-1.49$ for components #$1-3$, respectively. Throughout this paper, the ionization parameter ($\xi$) is given in units of ${\rm erg~{}s^{-1}~{}cm}$.

Two X-ray emission components with distinct velocity broadening are required for the soft X-ray emission lines in the 2002 and 2020 spectra (Figure 6 and Table 4). The warm emitter features are not prominent though, partly due to the relatively short exposure time. To reduce the number of free parameters, based on experience (e.g., Mao et al., 2018, 2019; Grafton-Waters et al., 2021), we fix the emission covering factor $C_{\rm em}=\Omega/4\pi$, where $\Omega$ is the solid angle subtended by the warm emitter with respect to the central engine. The emission covering factors of the two warm emitter components are 0.005 and 0.10, respectively. Both X-ray emission components are slightly (within $1-2\sigma$) broader and stronger in 2020. Nonetheless, the relatively short exposure for both X-ray spectra (Table 1) do not provide sufficient statistics to draw a firm conclusion.

The rising trend of the Swift X-ray hardness ratio since the summer of 2020 suggests an obscuration event, similar to those found in e.g., NGC 5548 (Kaastra et al., 2014) and NGC 3783 (Mehdipour et al., 2017). The apparent lowering of the soft X-ray flux in 2020, in comparison with that in 2002, supports the presence of an obscurer. We aim to constrain the physical properties of the obscurer with X-ray, UV, and NIR data.

The obscuration events in NGC 5548 (Kaastra et al., 2014; Mehdipour et al., 2016b, 2022a) and Mrk 335 (Longinotti et al., 2019; Parker et al., 2019) lasted for years, while relatively shorter and recurrent obscuration events were found in NGC 985 (Ebrero et al., 2016), NGC 3783 (Kaastra et al., 2018b), Mrk 817 (Kara et al., 2021), and NGC 3227 (e.g., Mehdipour et al., 2017; Wang et al., 2022; Mao et al., 2022). For short-lived obscuration events, the duration of the obscuration can be as short as 20 ks (for NGC 3227, Wang et al., 2022).

For MR 2251-178, if we consider the obscurer is present when the Swift X-ray hardness ratio is $\gtrsim 0.20$ (Figure 2), the obscuration duration is $\sim 75-260$ days. If the threshold is set to $\gtrsim 0.25$, the obscuration duration would be $\sim 75$ days.

Short-lived obscuration events are found to be recurrent for NGC 3783, NGC 985, and NGC 3227. For NGC 3783, the Swift hardness ratios between 2008 and 2017 suggest that there were obscuration events in early 2009 (Kaastra et al., 2018b) and December 2016 (Mehdipour et al., 2017, with joint UV and X-ray observations). Furthermore, archival X-ray spectra obtained with the Advanced Satellite for Cosmology and Astrophysics (ASCA) in 1993 and 1996 also show evidence of obscuration events (Kaastra et al., 2018b). For NGC 985, an obscurer was present with joint X-ray and UV observations in August 2013 and January 2015 (Ebrero et al., 2016). Archival X-ray data suggest that there might have been another obscuration event in July 2003 (Ebrero et al., 2016). For Mrk 817, obscuration has been persistent and variable from November 2020 through March 2021 (Kara et al., 2021), including a 55-day period similar to the “broad-line holiday” observed in NGC 5548. Moreover, an archival NuSTAR spectrum taken on 2015-07-25 suggests an earlier obscuration event, but less prominent than the late 2020 event (Kara et al., 2021). For NGC 3227, X-ray obscuration events were reported in $2000-2001$ (Lamer et al., 2003; Markowitz et al., 2014), 2002 (Markowitz et al., 2014), 2006 (Wang et al., 2022), 2008 (Beuchert et al., 2015), 2016 (Turner et al., 2018; Wang et al., 2022), and 2019 (Mao et al., 2022).

For PG 2112+059, obscuration events were observed on 2014-12-20 and 2015-08-29, but not on 2002-09-01 (Saez et al., 2021). With no observations between 2014-12-20 and 2015-08-29, it is hard to tell whether it was a continuous obscuration event lasting for eight months or two short-lived events.

For MR 2251-178, X-ray obscuration events might have occurred in 1980 (Halpern, 1984), 1996 (Markowitz et al., 2014), and late 2020 (present work). Compared to the Einstein X-ray spectrum of MR 2215-178 taken on 1979-07-01, the one on 1980-05-19 was significantly absorbed in the soft X-ray band below $\sim 3$ keV (Halpern, 1984). Markowitz et al. (2014) identified a clear obscuration event in the RXTE observation on 1996-12-09. In the two BeppoSAX observations in June and November 1998, X-ray obscuration had disappeared (Dadina, 2007). No other obscuration events were identified with the RXTE data up to January 2012 (Markowitz et al., 2014) and Swift data between 2014 and mid-2020 (Figure 1). With these X-ray observations and identified obscuration events, if there is a duty cycle, it would range from 16 to 24 years. Some periodic events might trigger the launch of the obscuring wind, or a warped accretion disk might cross our line-of-sight.

It is not trivial to estimate the distance of the obscurer to the black hole since we cannot directly resolve it via imaging. Lamer et al. (2003) estimated the distance of the obscurer in NGC 3227 assuming it is a spherical cloud orbiting the black hole in a Keplerian orbit. To be more specific, the distance is derived by (1) assuming the radius of the obscurer is $N_{\rm H}/n_{\rm H}$; (2) the obscurer crosses our line-of-sight with a constant velocity $v_{\rm cross}=\sqrt{GM_{\rm BH}/r}$ in a Keplerian orbit; (3) the crossing time $t_{\rm cross}=N_{\rm H}/(n_{\rm H}~{}v_{\rm cross})$ is constrained from the observed duration of the event. We estimate the distance of the 2020 obscurer in MR 2251-178 following the above approach but also discuss alternative scenarios below.

During the X-ray obscuration period of NGC 5548 (Kaastra et al., 2014; Mehdipour et al., 2016b), NGC 985 (Ebrero et al., 2016), NGC 3783 (Mehdipour et al., 2017; Kaastra et al., 2018b), Mrk 335 (Longinotti et al., 2019; Parker et al., 2019), and Mrk 817 (Kara et al., 2021), prominent blueshifted broad absorption troughs were found in the simultaneous HST/COS spectra. For the best-studied obscurer in NGC 5548, both its column density and covering fraction are variable on timescales of a few ks to a few months (Di Gesu et al., 2015; Mehdipour et al., 2016b; Cappi et al., 2016). The rapid variability and large velocity broadening support the scenario that the obscurer in NGC 5548 originates from the accretion disk (Kaastra et al., 2014). For NGC 3783, an additional high-ionization absorption component was also present in late 2016, leading to Fe xxv and Fe xxvi absorption lines in the Fe K band (Mehdipour et al., 2017). Through a detailed UV analysis, Kriss et al. (2019a) suggest that a collapse of the broad-line region (BLR) clouds triggered the launch of the obscurer. For Mrk 817, the obscurer is located at the inner BLR and partially covers the central source (Kara et al., 2021).

For NGC 3227, we do not find prominent blueshifted broad absorption troughs in the simultaneous HST/COS spectra when the target is obscured in the X-ray band (Mao et al., 2022). The lack of X-ray and UV association might be explained if the X-ray obscurer does not intercept our line-of-sight to (a significant portion of) the UV emitting region. If a compact X-ray obscurer intercepts our line-of-sight to the UV emitting region, it might cover $\lesssim 1$ % of the UV emitting region. It is also possible that the X-ray obscurer does not intercept our line-of-sight to the UV emitting region at all, although we cannot well constrain the distance of the obscurer.

For MR 2251-178, following Burke et al. (2021, Eq. S7), the effective UV emitting region radius ($R_{2500}$) can be estimated via

where $L_{5100}$ is the optical continuum luminosity. With $L_{5100}\sim 1.4\times 10^{44}~{}{\rm erg~{}s^{-1}}$, we obtain $R_{\rm 2500}\sim 1.1\times 10^{15}~{}{\rm cm}$ or $\sim 0.41$ ld. This is equivalent to $\sim 18~{}R_{\rm S}$, where the Schwarzschild radius $R_{\rm S}=2GM_{\rm BH}/c^{2}=5.9\times 10^{13}~{}{\rm cm}$ or $0.023$ ld.

Assuming a fiducial X-ray emitting central engine with a size scale of $\lesssim 5~{}R_{\rm S}$ (e.g., Reis & Miller, 2013; Fabian et al., 2015), the UV emitting region would be a factor of $\gtrsim 13$ larger. With the X-ray obscurer covering $\sim 35$ % of the X-ray emitting central engine (Table 5), it would then cover merely $\lesssim 2.7$ % of the UV emitting region.

The X-ray covering factor ($f_{\rm cov}^{X}$) of MR 2251-178 is comparable to that reported by Ebrero et al. (2016) for the observation episode in NGC 985 observed with XMM-Newton in January 2015. In that case, only one narrow (FHWM$\sim 350\pm 50~{}{\rm km~{}s^{-1}}$) blueshifted ($-5300\pm 10~{}{\rm km~{}s^{-1}}$) absorption line of Ly$\alpha$ was present in the simultaneous HST/COS spectrum. The UV covering factors for the low er-ionization ions C iii*, C iv, and Si iv were $\lesssim 0.15$ and not visible at the signal to noise ratio of the UV spectra.

On one hand, if the obscuring wind barely intercepts the UV continuum region (e.g., the wind projected size is too small compared to the UV emitting region), we might not observe prominent blueshifted absorption lines in the UV band. On the other hand, when a transient obscuring wind is fading, we might not observe prominent blueshifted absorption lines in the UV band. Both might contribute to the lack of prominent UV absorption lines in MR 2251-178.