Integral Field Spectroscopy of the inner kpc of the elliptical galaxy NGC 5044

TL;DR

This study uses integral field spectroscopy to analyze the stellar populations, gas kinematics, and ionization mechanisms in the inner kiloparsec of NGC 5044, revealing a weak AGN, a rotating gas disk, and filamentary inflows.

Contribution

First detailed integral field spectroscopic analysis of NGC 5044's nuclear region, identifying a broad Hα component and gas inflows fueling the AGN.

Findings

Presence of a broad Hα emission line in the nucleus.

Detection of a rotating gas disk with a velocity of ~240 km/s.

Mass inflow rate of ~0.4 solar masses per year sufficient to power the AGN.

Abstract

We used Gemini Multi-Object Spectrograph (GMOS) in the Integral Field Unit mode to map the stellar population, emission line flux distributions and gas kinematics in the inner kpc of NGC 5044. From the stellar populations synthesis we found that the continuum emission is dominated by old high metallicity stars ($\sim$13 Gyr, 2.5Z$\odot$). Also, its nuclear emission is diluted by a non thermal emission, which we attribute to the presence of a weak active galactic nuclei (AGN). In addition, we report for the first time a broad component (FWHM$\sim$ 3000km$s^{-1}$) in the H$\alpha$ emission line in the nuclear region of NGC 5044. By using emission line ratio diagnostic diagrams we found that two dominant ionization processes coexist, while the nuclear region (inner 200 pc) is ionized by a low luminosity AGN, the filamentary structures are consistent with being excited by shocks. The H$\alpha$ velocity field shows evidence of a rotating disk, which has a velocity amplitude of $\sim$240kms$^{-1}$ at $\sim$ 136 pc from the nucleus. Assuming a Keplerian approach we estimated that the mass inside this radius is $1.9\times10^9$ $M_{\odot}$, which is in agreement with the value obtained through the M-$\sigma$ relation, $ M_{SMBH}=1.8\pm1.6\times10^{9}M_{\odot}$. Modelling the ionized gas velocity field by a rotating disk component plus inflows towards the nucleus along filamentary structures, we obtain a mass inflow rate of $\sim$0.4 M$_\odot$. This inflow rate is enough to power the central AGN in NGC 5044.

Tables

Table 1. Table 1: Basic parameters of NGC 5044.

Parameter	NGC 5044
RA (2000)	13h 15m 24.0s
DEC	-16∘ 23’ 09”
Morphology	E0
B (mag)	11.83
MB (mag)	-22,39
Av (mag)	0.192
Radial velocitya (km s-1)	2782
Distance (Mpc)	37
Redshift (z)	0.009280
Lx b (erg s-1)	6.8e+39
Data available in NED111The NASA/IPAC Extragalactic Database (NED) is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.
a Radial velocity measured by Ogando et al. (2008).
b X-ray luminosity in 0.3–8 keV (Liu, 2011).

Table 2. Table 2: Observation log.

Parameter	NGC 5044
Observation date	May 04 2013
Gemini Programme	GS-2013A-Q-52
Seeing (arcsec)	$\sim$ 0.7
Airmass	1.14
Texp(s)	1800

Table 3. Table 3: Results for the Gaussian profile fits measured for H α 𝛼 \alpha emission line in selected regions.

Region	Set	FWHM ($\mathrm{km}{s}^{-1}$)	$V_c$ ($\mathrm{km}{s}^{-1}$)	$\left|\Delta V\right|$($\mathrm{km}{s}^{-1}$)
1	1	$215\pm 28$	$-178\pm 34$	$196\pm 50$
	2	$172\pm 25$	$18\pm 37$
2	1	$376\pm 40$	$46\pm 25$	$407\pm 37$
	2	$301\pm 35$	$-361\pm 28$
3	1	$323\pm 36$	$-119\pm 28$	$283\pm 44$
	2	$194\pm 27$	$-402\pm 35$
4	1	$376\pm 40$	$-23\pm 25$	$274\pm 39$
	2	$280\pm 33$	$-297\pm 30$
Nuc	Narrow	$1076\pm 68$	$5\pm 15$	$558\pm 17$
	Broad	$3001\pm 113$	$562\pm 9$

Table 4. Table 4: Gas velocity field parameters.

Parameter	Values
A (km s-1)	$1285\pm 189$
$c_0$ (pc)	$95\pm 5$
p	$1.5\pm 0.1$
vs (km s-1)	$2782\pm 1$
${\psi }_0$	$152°\pm 2°$
$\Delta x$ (pc)	$0\pm 1$
$\Delta y$ (pc)	$18\pm 1$

Equations

$$M_{\lambda }=M_{\lambda 0}\left[\sum _{j=1}^{N_{\star }}{x}_j{b}_{j,\lambda }{r}_{\lambda }\right]\otimes G(v_{\star },{\sigma }_{\star }),$$

$$\text{centering}{\delta }_{v_r}=\frac{c}{{\lambda }_0}{\delta }_{{\lambda }_0};{\delta }_{{\lambda }_0}=\frac{I_0\sqrt{N}}{F_{\lambda }}{\delta }_{\lambda },\text{@add@centering}$$

$$\mathrm{V}(\mathrm{R},\psi )={\mathrm{V}}_{\mathrm{s}}+\frac{\mathrm{ARcos}(\psi -{\psi }_0)\sin \theta \mathrm{co}{\mathrm{s}}^{\mathrm{p}}\theta }{{\mathrm{R}}^2[\mathrm{si}{\mathrm{n}}^2(\psi -{\psi }_0)+\mathrm{co}{\mathrm{s}}^2\theta \mathrm{co}{\mathrm{s}}^2(\psi -{\psi }_0)]+{\mathrm{c}}_0^2\mathrm{co}{\mathrm{s}}^2{\theta }^{\frac{\mathrm{p}}{2}}}.$$

$$M_{bh}=0.309_{-0.033}^{+0.037}\times 10^9{M}_{\odot }{\left(\frac{{\sigma }_{\star }}{200km{\mathrm{s}}^{-1}}\right)}^{4.38\pm 0.29},$$

$$\text{centering}{M}_{HII}=\frac{L_{H\alpha }}{h{\nu }_{H\alpha }}\frac{m_p}{n_e{\alpha }_{H\alpha }^{eff}},\text{@add@centering}$$

$$\text{centering}{\dot{M}}_{inflow}=4N_{HII}\upsilon A,\text{@add@centering}$$

$$\text{centering}\dot{M}=L_{bol}/(\eta c^2),\text{@add@centering}$$

Full text

††pagerange: Integral Field Spectroscopy of the inner kpc of the elliptical galaxy NGC 5044–LABEL:lastpage††pubyear: 2016

1 INTRODUCTION

It is known for a long time that many elliptical galaxies show disks, as well as filaments of ionized gas (Macchetto et al., 1996; Ferrari et al., 1999; Tran, 2001). Many of the early type emission line galaxies show characteristic spectra of a low-ionization nuclear emission-line region (LINER, Heckman, 1980). However, the ionization mechanisms of this gas is still matter of debate. For example, very recently, Ricci et al. (2015) studied early-type galaxies with ionized gas using optical Integral Field Spectroscopy (IFS) and found that the H$\alpha$ flux along the disk cannot be explained only by ionizing photons emitted by an Active Galactic Nuclei (AGN). Other mechanisms, like ionization by shock (Heckman, 1980; Dopita & Sutherland, 1995; Allen et al., 2008; Dopita et al., 2015) or Post Asymptotic Giant Branch (Post-AGB) stars (Stasinska et al., 2008; Singh et al., 2013), can also explain the observed spectra. This may suggest that the gas is being ionized by more than one ionization mechanism.

Furthermore, large amount of molecular gas (up to 10${}^{7}\rm M\odot$ revelled by CO(2-1) 1.3 mm and H2 2.121 $\mu$m emission) was found for a sample of bulge-dominated galaxies with dust lanes, revelling the complexity of these galaxies (David et al., 2014; Davis et al., 2015; Strong et al., 1988). It was suggested that this molecular and atomic gas would be accreted through a recent merger with a gas rich galaxy and/or that it originates from a cooling flow (O’Sullivan et al., 2015; David et al., 2014). In addition, interstellar dust, mixed with the molecular and atomic gas is observed in early type galaxies (e.g. Davis et al., 2015, and references therein). Disks and dust filaments at kpc scales (with mass between 10 and 200 ${M}_{\odot}$ and Mid-Infrared luminosity up to 14.6 $\times$ 108 ${L}_{\odot}$) following the ionized gas distribution were also reported in early type galaxies (Ferrari et al., 1999, 2002; Goudfrooij et al., 1994).

Nowadays, galaxies with LINER-like spectrum are defined based on the optical line ratios (Ho, 2008): [OI]$\lambda$6300 / [OIII]$\lambda$5007 $\geq$ 1/3, [NII]$\lambda$6583 / H$\alpha$ $\geq$ 0.6, [OIII]$\lambda$5007 / H$\beta$ $\leq$ 3. In addition, over the last 30 years the most accepted explanation was that these objects are powered by weak AGNs (Ferland & Netzer, 1983; Halpern & Steiner, 1983). However, in the last few years contradictory results were found between the AGN-ionization scenario and predicted emission line strengths (Cid Fernandes et al., 2011) or with the spatial distribution of the ionized gas regions (e.g. Sarzi et al., 2010; Yan & Blanton, 2012; Martins et al., 2013; Belfiore et al., 2015). In addition, using the Calar Alto Legacy Integral Field Area (CALIFA, Sanchez et al., 2012) survey of spatially resolved data Singh et al. (2013) have shown that galaxies with LINER-like spectrum are powered by hot old stars and accreting black holes are rare. In this work we propose to test this hypothesis by mapping in details the emission gas and stellar populations in the inner kiloparsec in a key object – the gas rich early type galaxy NGC 5044.

NGC 5044 was observed over almost the entire electromagnetic spectrum, presenting rich structures, such as gas and dust filaments (Ferrari et al., 1999). It presents a very bright ionized gas emission in form of filaments extending up to 10 kpc with a complex kinematics (Macchetto et al., 1996). Its radial velocity is blueshifted with respect to the systemic stellar velocity and the inner parts of the stellar velocity curves presents counter-rotation with respect to the galaxy outer regions (Caon et al., 2000).

The gas ionization mechanism of NGC5044 has been the aim of several studies, for example, Rickes et al. (2004) suggest that the gas in the central region ($\sim$4$\arcsec$) of NGC 5044 is ionized by a Low Luminosity AGN (LLAGN), while the outer regions are explained by post-AGB stars (see also Macchetto et al., 1996). In addition, the overall stellar emission of this galaxy seems to be dominated by old stars over a large wavelength range, from UV (Annibali et al., 2007; Marino et al., 2011), optical (Rickes et al., 2004) and Mid-infrared observations (Vega et al., 2010). Unusual polycyclic aromatic hydrocarbons (PAH) emission was detected in Spitzer-IRS observations by Vega et al. (2010), this emission was not expected in this galaxy because PAH is a typical feature in star-forming galaxies. Moreover, the IRS spectrum of this object displays strong H2 molecular and atomic gas emission (Vega et al., 2010; Panuzzo et al., 2011). Furthermore, this galaxy shows several X-ray cavities within the inner 10 kpc, which are attributed to multiple AGN outbursts over past 108 yrs (David et al., 2009; Gastaldello et al., 2009). Moreover, David et al. (2014) using Atacama Large Millimeter/submillimeter Array (ALMA) data argue that the detection of 230 GHz continuum emission shows that it is currently going through another outburst, probably due a recent accretion event. In addition, these authors report the detection of cold molecular CO gas moving towards the center of this galaxy, and suggested that it can be interpreted as the presence of a cooling flow in the galaxy center direction.

We present in this paper for the first time the optical IFS obtained with Gemini Multi-Object Spectrograph (GMOS) in integral field unit (IFU) of the central 1 kpc of NGC 5044, whose gas ionization source is still controversial. Beside that, there is no information about its central gas kinematics. Although strong evidence suggests possible presence of a contribution of an AGN, other energy source also seem to be present in this galaxy ionizing their extended filaments (Macchetto et al., 1996; Rickes et al., 2004). This data allowed us to spatially resolve the inner kpc of this source and to map the possible ionization mechanisms through the analysis of stellar populations and emission line flux ratios. The goals of this work are to establish the star formation history (SFH), the nature of the energy sources that ionizes the gas and search for the existence of inflowing or outflowing gas associated with the AGN and/or other mechanism, like cooling flows, for the central kpc of NGC 5044.

This paper is structured as follows: In Sec. 2 we described the observations and data reduction procedures. Details of the spectral synthesis method and its results are presented in Sec. 3. In Sec. 4 we analyse the gas kinematics. The emission gas and its ionization mechanism is studied in Sec. 5. We show the velocity field model and estimates the black hole mass in Sec. 6. The geometry and the origin of the observed gas is discussed in Sec. 7. Lastly, we present our conclusions in Sec. 8.

2 OBSERVATIONS AND DATA REDUCTIONS

NGC 5044 was observed on May 04, 2013 with the Gemini South Telescope using the GMOS in the IFU mode (Allington-Smith et al., 2002; Hook et al., 2004) using the one slit setup. In this configuration the spectrograph uses 500 hexagonal micro-lenses centered on the object, and other 250 micro-lenses, separated by 1 $arcmin$ from the object, to simultaneously measure the sky emission. These micro-lenses, located at the focal plane of the telescope, divide the image in slices with a diameter of 0.2$\arcsec$ and are connected to a set of fibers. This setup produces a data cube with one spectral dimension and two spatial dimensions with a field-of-view of 3.5 x 5 $arcsec^{2}$. We used the B600-G5323 grating, with a central wavelength of 5620$\AA$. The spectra covered a range from 4260 to 6795 $\AA$ with a 1.7$\AA$ resolution, obtained from the full width at half maximum (FWHM) of CuAr arc lamp lines. In Table 1 we list some of the general parameters of NGC 5044. Flat-Field exposures, bias and CuAr lamp spectra were obtained for calibration and correction of the data cubes. The DA white dwarf EG 274 (Hamuy et al., 1992) was used to perform the spectrophotometric calibration. The seeing value of the observation was measured in the acquisition r image obtained with the GMOS imager (SDSS system). This value and other observation parameters are shown in Table 2.

We followed standard procedures to reduce the data, using iraf (Tody, 1986, 1993) and the tasks contained in the gemini iraf package, as described in detail in Ricci et al. (2014). The data cube was built with a spatial sampling of 0.05$\arcsec$, resulting in a FoV with 66$\times$98 spaxels. The high spatial frequency noise was removed with a butterworth filter (Gonzalez & Woods, 2002; Ricci et al., 2014) using order n=6 and a cut-off frequency Fc=0.13 FNy, where FNy = 0.5 spaxel*-1* is the Nyquist frequency. This cut-off frequency was chosen to remove only spatial frequencies higher than the seeing. In addition, the Principal Component Analysis (PCA) tomography (Heyer & Schloerb, 1997; Steiner et al., 2009) was applied in order to remove signatures of low spatial and spectral frequency (Ricci et al., 2014). To correct the atmospheric refraction effects, we applied the equations proposed by Bönsch & Potulski (1990) and Filippenko et al. (1982).

The data cube spectra was corrected for Galactic extinction using the Cardelli, Clayton & Mathis (1989) extinction curve (CCM) for an $A_{v}=0.192$ (Schlafly & Finkbeiner, 2011) and by Doppler effect using the radial velocity of 2782 $\rm km\,s^{-1}$given by Ogando et al. (2008). These authors derived the radial velocity through the cross-correlating technique by using the RVSAO package (Kurtz & MinK, 1998) with stellar spectra of G and K giant stars.

We show in the left panel of the Figure 1 a H$\alpha$+[NII] emission image of NGC 5044 obtained with NTT+EFOSC2 (Macchetto et al., 1996), where the black rectangle in the middle represents the GMOS FoV. In the right panel we presented two images extracted from our data cube. The continuum image was obtained by estimating the mean value over the wavelength range between 4730 and 4780 $\AA$, with the continuum peak defined as the photometric center of the galaxy. We also show the H$\alpha$ image with contours drawn to highlight the extended ionized gas filaments. It is clear from these maps that the morphology of the continuum and H$\alpha$ are distinct and suggests that the ionized gas follows the larger kpc scale filamentary structure, while the continuum follows the stellar distribution.

The wavelength calibration uncertainty was estimated from the arc lamp lines (see section 4). The spectroscopic standard star was not observed at the same night as the galaxy and this may be affected the accuracy on the absolute flux determination, given 30% of uncertainty in their values. It is worth mentioning that the emission line flux ratios, used to build the diagnostic diagrams were not affected by the systematic flux uncertainty, since each used ratio were measured over nearby emission lines. Nevertheless, we check uncertainties for [NII]$\lambda$6584/H$\alpha\lambda$6563 and [OIII]$\lambda$5007/H$\beta\lambda$4861 ratios in several spaxels and found a deviation of 3%. The instrumental dispersion was obtained from the FWHM of the CuAr arc lamp, and we used it to corrected the velocity dispersion. We noted that the deviation are larger in blueward than in redward, given an uncertainties of $\sim$8% and $\sim$4% respectively.

3 Stellar Population Distribution

In order to obtain the SFH and to analyse the gas emission free from the stellar population contamination, we performed stellar population synthesis spaxel by spaxel. Therefore, we use the starlight code (Cid Fernandes et al., 2004, 2005; Mateus et al., 2006; Asari et al., 2007; Cid Fernandes et al., 2009), which essentially fits the whole underlying observed spectrum, $O_{\lambda}$, with a model spectrum, $M_{\lambda}$. To create a model spectrum, $M_{\lambda}$, the code finds a combination of $N_{\star}$ simple stellar populations (SSPs) resulting in a final population vector x, whose components represent the fractional contribution of each SSP to the total synthetic flux at wavelength $\lambda_{0}$ (Cid Fernandes et al., 2004, 2005). Extinction is parametrised by the $V$-band extinction $A_{V}$ and modelled by starlight as due to foreground dust. In summary the code solves the following equation:

[TABLE]

where $M_{\lambda 0}$ is the synthetic flux at the normalisation wavelength, $G(v_{\star},{\sigma}_{\star})$ is a Gaussian function used to model the line-of-sight stellar velocity distribution, which is centered at velocity $v_{\star}$ with dispersion $\sigma_{\star}$. The term $x_{j}$ is the jth population vector component of the base set; $b_{j,\lambda}$. The spectra of the SSP of the base of elements are normalized at $\lambda_{0}$, the reddening term is represented by $r_{\lambda}=10^{-0.4(A_{\lambda}-A_{\lambda 0})}$. The final fit is carried out by minimizing an $\chi^{2}$ equation. The robustness of the fits can be measured by the starlight output parameter ADEV, which is the average deviation over the fit of pixels $|O_{\lambda}-M_{\lambda}|/O_{\lambda}$.

It is clear from the above that the most important ingredient in stellar population synthesis are the SSPs used in the fit (what we call base set). We used SSPs taken from Bruzual & Charlot (2003) models. These models were chosen because they are the most adequate to fit our data, since they have a spectral resolution of 3.0 Å, that is close to our data spectral resolution (see section 4) and do offer enough spectral coverage (3200 - 9500Å) to fit our data. Beside they do provide an adequate number of SSP in order to have all the possibilities of ages and metallicities found in galaxies. In addition, we decided to use the BC03 models because they are widely used, thus allowing for comparison with published results. These models are computed using the isochrone synthesis code developed by BC03 and are divided in 221 spectra with ages between 0.1 Myr - 20 Gyr for a wide range of metallicities. These models mainly use the BaSeL 3.1/STELIB libraries, with a spectral resolution of 3Å (STELIB) in the optical region. Following Dametto et al. (2014), our final base set was selected in order to have enough, representative and not degenerated elements, covering 7 ages ($t=$ 0.6, 0.9, 1.4, 2.5, 5, 11 and 13 Gyr) and three metallicities (0.2  Z$\odot$, 1  Z$\odot$, 2.5  Z$\odot$). In the fits the spectral resolution of the data was resampled to match the models. The Cardelli, Clayton & Mathis (1989) extinction law, which is typically used to fit the extinction data both for diffuse and dense interstellar medium. Since we are fitting a LINER-like spectrum galaxy, we additionally included a featureless continuum (FC) in the form of a power law ($F_{\nu}\propto\nu^{-1.5}$) in order to account for a possible non thermal contribution (e.g. Cid Fernandes et al., 2005; Riffel et al., 2009, and references therein).

To map the whole FoV, the stellar population fits were performed for each one of the spaxels. The nuclear region ($\sim$200 pc) was defined as the one where a broad component was detected on H$\alpha$ emission. Fig. 2 shows the fit of the spectrum at the nuclear spaxel, defined as the location of the continuum peak. The emission lines and the final continuum fit are also shown. The percentage contribution of each age (in mass222The mass fractions are obtained using the M/L ratios for each SSP, see starlight manual for details and light fractions) and FC to the final fit are presented in the bottom panels. Here we discuss only the light fractions, since it is the parameter we derive directly and where larger differences are observed (in terms of mass, the emission is completely dominated by the old stellar population which has an high M/L).

The quality of the fit over the whole FoV can be seen in Fig. 3 where the parameters ADEV and $\chi^{2}$, are shown. In this figure we do also show a signal to noise ratio (SNR) map. It is clear from this map that even for the borders of the FoV a SNR $>$15 is found. It is worth mentioning that according to tests presented by Cid Fernandes et al. (2004), for spectra with values of SNR $>$ 10 starlight does properly reproduce the input parameters, thus producing reliable results for the stellar populations fits. However, the strengths of the Mg$\lambda$5143 and Na$\lambda$5894 lines are not well reproduced by our fitting, possible due to the fact that the BC03 models do not consider $\alpha$ enhancement processes. In addition, the NaD doublet has contribution of the ISM. Due to this fact we have tested the stellar population fits in three cases, with and without removing both the NaD and Mg$\lambda$5143 lines form the fit and only removing the NaD doublet. We have not found any significant difference between both fits. In both cases, less than $\sim$ 10% of these lines strength was not reproduced by the fitted model. Likely the NaD residual seen in Figure  2, is due to ISM contribution present in the nucleus of NGC 5044 (see Ferrari et al., 1999, for example). As discussed in Davis et al. (2012) the blueshifted NaD doublet provides a good probe of the cold interstellar medium outflow. In the case of NGC 5044 we have not found any blue-shifted component for NaD doubled in all our FoV, indicating that in this galaxy, the cold material is probably not outflowing.

Interestingly, our fitting resulted only in contributions of four of the base components: three SSPs with ages 900 Myr (0.2 $Z\odot$) and 13 Gyrs (0.2 and 2.5 $Z\odot$), plus a FC component. The 2D maps for these components (in light fractions), together with mean age, Av, FC and mean metallicity are shown in Fig. 4. It is clear from this figure that the light in the central region of NGC 5044 is dominated by an old stellar population. Besides, it is also clear that an intermediate age (900 Myr) stellar population is present, surrounding the galaxy nucleus. As far as we know, the presence of stellar populations with these ages (900 Myr) in the nuclear region of elliptical galaxies is not a common trend. For example, from an extensive stellar population study performed with large field data of the CALIFA survey, González-Delgado et al. (2015), found that in average the S0 and E galaxies have a dominant stellar population with almost the same age and they are older than those found in spirals. In addition, they reported for elliptical galaxies a small variation on the stellar ages along the galaxy radii (from 1010 to 109.7 yr from the center to 3 Half Light Radius, (see figure 17 of González-Delgado et al., 2015). The differences in the results found by González-Delgado et al. (2015) and ours may be due to the spatial and spectral resolution probed by GEMINI GMOS. For instance, the presence of spatially resolved circumnuclear intermediate age stellar populations, like we found in NGC5044, was reported for the bulges of spiral Seyfert galaxies (Riffel et al., 2010, 2011; Storchi-Bergmann et al., 2012). The maps of mean age and mean metallicities shows a trend consistent with the stellar population becoming more metal rich and older with decreasing distance with respect to the galaxy centre. Some studies have shown that the stellar population in early type galaxies also become progressively more metal rich and older with increasing velocity dispersion (Thomas et al., 2005; Clemens et al., 2006, 2009). Indeed, we found higher $\sigma$ values in the central region of the NGC 5044 (see Sec.  4), thus our result does further suport the findings by these authors.

In addition, our stellar population results are in agreement with the results found by Rickes et al. (2004), who performed stellar population synthesis using optical long slit data in a much larger spatial scale than ours. These authors used two different bases of elements, one with three components with metallicities [Z/$Z_{\odot}$] = 0, -0.4 and -1.1 and age = $10^{10}$ years, and another with solar metallicity and ages with $10^{10}$yrs, $10^{8}$yrs and 106yrs. They noted that there was an improvement on the fit when considering different metallicities over the different ages, with the more metallic component dominating in the central region, while the lower metallicities were enhanced in the outer regions.

Literature results on the stellar content of NGC 5044, as those presented by Werner et al. (2014) based on the GALEX and WISE data report an unusual star formation rate in NGC 5044 (SFR= 0.073 $M_{\odot}/yr$, using a region of 10 kpc, over the last 1 Gyr). Besides, David et al. (2014) suggested that the unusual lines of PAHs found in NGC 5044 and the uncertainty in the stellar age are consequences of episodes of star formation that occurred in the last 1 Gyr (Marino et al., 2011). These results are further supported by our finding, that there is a significant contribution of the 900 Myr stellar population, distributed along the center of the galaxy (reaching values of $\sim$ 50% at some locations).

We also found a FC contribution of about 20% in the nuclear region. This component is consistent with a scenario where the gas in the nuclear region is ionized by a weak AGN. However, we find some significant contribution of the FC in the external regions. Since this component is degenerated with a young reddened stellar population we interpret this contribution as being related with the fact that BC03 models do not include blue horizontal branch stars, and starlight fits tend to choose a young population to account for this blue population (see Cid Fernandes & Gonzalez-Delgado, 2010, for details).

The visual extinction map (Av) shows a clear enhancement at the galactic centre, reaching $\sim$0.35 mag and decreasing to 0.1 mag at the external region. Ferrari et al. (1999) found a mean extinction Av=0.01 (mag) within a region of $10^{\prime\prime}$, which is about twice of the size of our region. From this Av, they estimated a dust mass of about 104M*⊙* distributed irregularly along the galaxy. In addition, a hot dust mass of 67 M*⊙* was determined by Ferrari et al. (2002) from Mid-IR data.

4 kinematics of ionized gas

In order to map the gas kinematics we measured the centroid velocity (hereafter velocity field, ${v_{r}}$) and velocity dispersion ($\sigma_{v_{r}}$) from the data cube without the subtraction of the stellar populations contribution (see Sec. 3), because the stellar population templates have slightly lower resolution than the data cube, this difference would affect the accuracy on velocities determination. We have used an in house fitting code (Hernandez-Jimenez et al. - in preparation) to fit Gaussian functions to the emission-line profiles. We fitted single- or multi-Gaussian components to the spectra in order to obtain the flux of each emission line. The line pseudo-continuum is linearly fitted using side-bands regions, free from emission or absorption features, as close as possible to the emission line. The uncertainty associated to the measurements of the velocity field ($\delta_{v_{r}}$), in a given line at $\lambda_{0}$, depends implicitly on the error of the Gaussian peak centroid, ($\delta_{\lambda_{0}}$). Then, we do a straightforward error propagation of the basic expression for the Gaussian peak centroid ($\lambda_{0}$) to calculate $\delta_{v_{r}}$. We derived the following expression:

[TABLE]

where $I_{0}$ is the intensity at maximum of the fitted Gaussian, $N$ is the number of pixels, $F_{\lambda}$ is the line flux, and $\delta_{\lambda}$ is the error of the spectral dispersion taken from the FWHM (1.7Å) of the CuAr arc lamp. The error map $\delta_{v_{r}}$ of H$\alpha$ line is shown in panel $(d)$ of Fig. 10, where we can see that the uncertainties in velocity field in the nuclear region is lower than 10 $\rm km\,s^{-1}$, while at the data cube border (where the SNR is lower), the error is $\sim$ 20 $\rm km\,s^{-1}$.

In order to study the kinematic components of the ionized gas in the circumnuclear region of NGC 5044, we obtained channel maps between -431 $\rm km\,s^{-1}$ and +489 $\rm km\,s^{-1}$  relative to the systemic velocity of the galaxy derived by modelling the gas velocity field (see Sec. 5 and Table 4), in velocity bins of 80 $\rm km\,s^{-1}$  (see Figure 5). We identified, in the negative channel map -180 $\rm km\,s^{-1}$, a bright structure at NE, named as region 1, and a weak structure (region 2) at west, which is also observed at the channel -96 $\rm km\,s^{-1}$. In the positive channel maps the most conspicuous structure was named as region 3 and is located at South, at the channel 70 $\rm km\,s^{-1}$. There is also a filamentary structure, region 4, located at North between 405 and 489 km s*-1*. Comparing these ionized gas structures with the ones of molecular gas identified by David et al. (2014), using ALMA observations, we found that the brighter peak located at NE of the channel map -180 km s*-1* match with the approaching cloud 13 (see Figure 6 in David et al., 2014), whereas the south filament in channel map 70 km s*-1* coincides with the receding molecular gas cloud 18. In both regions David et al. (2014) found evidence that the molecular gas is falling into the center of the galaxy. On the other hand, we point out that the central region appears in all channel maps, indicating the presence of a nuclear broad component with a FWHM $\sim$ 3000 km s*-1*, typical of AGN.

We fitted the [NII]+H$\alpha$ complex at the nuclear region with six gaussians, five corresponding to the narrow components of the lines [NII]$\lambda$6548, H$\alpha\lambda$6563, [NII]$\lambda$6583, [SII]$\lambda\lambda$6717,31 and one to account for the H$\alpha$ broad component. We constrained the narrow components as having the same redshift and FWHM for all lines, and fixed [NII]$\lambda$6583/[NII]$\lambda$6548 = 3.1 (Osterbrock & de Robertis, 1985). The fit for the additional component was done with all parameters free, resulting in a large FWHM value ($\sim$3000 $\rm km\,s^{-1}$), thus we associated it with a broad H$\alpha$ emission. To properly reproduce the emission line profiles of the outer regions (1, 2, 3 and 4), it was necessary two sets of gaussians, constrained as mentioned above, but each set with their own values for redshift and FWHM. We only considered lines with SNR above than 3$\sigma$ of the fitted continuum (see Fig. 6). In Table 3 we show the values obtained for each fit, where ${V_{c}}$ is the peak velocity of the line, and $|\Delta V|$ is the shift between the line sets. The fitted H$\alpha$+[NII] and the [SII] lines profiles, obtained with apertures of the 0.7″, are shown in Figure 6. The secondary set of lines of regions 2, 3 and 4 are blueshifted with respect to the first one. The difference in velocity between the sets for region 2 is 407 $\rm km\,s^{-1}$(the largest one). We point out that this region is the closest from the nucleus. The velocity difference for regions 3 and 4, are $\sim$300 $\rm km\,s^{-1}$. The secondary set of lines for region 1 is redshifted with respect to the primary one in about 200 $\rm km\,s^{-1}$. Figure 6 also shows the line profiles for the nuclear region, where the H$\alpha$ broad component is blueshifted with respect to the narrow component by 558 $\rm km\,s^{-1}$. The FWHM of these components are $\sim$1080$\rm km\,s^{-1}$and $\sim$3000 $\rm km\,s^{-1}$, respectively. It is clear from Figure 6, that the regions 2 and 4 do present a wider FWHM than regions 1 and 3. Besides they do show a weak red wing in the [NII]+H$\alpha$ complex (very close to the 3$\sigma$ detection limit). We have not a very clear explanation for this component, however, we do speculate that this weak component is most likely originated from material outflowing from the AGN, once these two locations are closer to the galaxy center. Similar results were found for other early type galaxies, for example, Ricci et al. (2014) fitted two sets of Gaussians to the emission lines in the inner 100 pc of 10 early type galaxies and found that for seven galaxies one of these sets has FWHM $>$390 $\rm km\,s^{-1}$and the other (which corresponds to the H$\alpha$ broad component) has FWHM $>$2000 $\rm km\,s^{-1}$. In summary, this exercise indicates that there are three kinematical components which we discuss in the following sections.

Velocity field and dispersion maps measured on H$\alpha$ are shown in Figure 10. They were obtained from a fit of a single component for each emission line over the whole FoV, except for the nuclear region where a very broad component was added. Analysing the velocity field map, we found evidence of rotating gas and that the velocity field is asymmetric333This asymmetry can be due to the non-rotational motions of the filamentary structure observed in this galaxy, see section 7., with the receding side at the south and approaching side at the north. The maximum approaching value is -120 $\rm km\,s^{-1}$at $\sim$200 pc and the receding one is $\sim$200 $\rm km\,s^{-1}$at $\sim$100 pc. The line-of-nodes lies in PA$\sim$ 152$\,{}^{\circ}$.

The H$\alpha$ velocity dispersion map corrected by the instrumental broadening ($\sigma$, Figure 10e), shows a large range of values (with a constant mean uncertainty of 4$\rm km\,s^{-1}$) of through the overall FoV, from 100 $\rm km\,s^{-1}$up to 600 km s*-1*, the peak of the distribution has an offset of $\sim$ 30 pc to SE with respect to the photometric center, marked with a symbol “x” in all maps. Two different regions can be clearly distinguished from the distribution of $\sigma$ values. The first one is the nuclear region that shows a $\sigma$ gradient from 300 $\rm km\,s^{-1}$to 600 $\rm km\,s^{-1}$. The second region presents a “dove-like” distribution of $\sigma$ values between 150 and 300 km s*-1*, with “wings” on the east and north directions and the “body” in the SW direction. We noted that, in the rest of the FoV, the velocity field and dispersion velocity have similar values ($\sim$100 $\rm km\,s^{-1}$).

5 distribution and ionization mechanisms of the ionized gas

The results of the individual fits obtained from the stellar synthesis can be subtracted spaxel-by-spaxel, letting us with a clean residual data cube, free from the stellar population absorption features (see bottom Figure  2), thus, allowing us to study the pure emission gas component. To measure the emission line fluxes we used the same methodology described in Sec. 4.

Figure 7 present the ionized gas flux distribution for following emission lines: H$\alpha$ total flux, [NII]$\lambda$6583, [SII]$\lambda$6717+$\lambda$6731 and [OIII]$\lambda$5007. In the inner 1 kpc of the NGC 5044, the gas distribution is irregular. The flux maps for the emission lines [NII] and [SII] show that their emission are more concentrated at the nucleus and indicate a preference of distribution along south-north direction, while the [OIII] map shows emission over the whole FoV, with regions of higher intensity extended along the E-W direction. Also, the flux distribution for other emission lines detected in our data (H$\beta$, [OIII]$\lambda$4959 and [OI]$\lambda$6300), do have a lower SNR and are shown for completeness.

From the flux maps, we derived the following values for the emission line ratios: 0.1 $<$ log ([NII]$\lambda$6583/H$\alpha$) $<$ 0.5, -0.05 $<$ log [SII]$\lambda\lambda$/H$\alpha$ $<$ 0.5 and -0.4 $<$ log [OIII]$\lambda$5007/H$\beta$ $<$ 0.5, which are typical of LINER-like emission line objects (Heckman, 1980; Ho, 2008).

The dominant ionization/excitation mechanism of the ionized gas in NGC 5044 is still controversial. According to Macchetto et al. (1996) the observed spectra can be explained by post-AGB stars, while Rickes et al. (2004) suggest that the spectra have contributions of a LLAGN in the nuclear region plus post-AGB stars in the outer parts.

In order to study the dominant ionization mechanism of the emission line gas in NGC 5044, we constructed diagnostic diagrams using the emission line ratios mentioned above. The diagnostic diagrams for the nuclear region and for the regions 1 to 4 are shown in Figures 8 and 9, respectively.

In the diagnostic diagrams, in order to discriminate among the different excitation mechanism, we added the theoretical maximum starburst line of Kewley et al. (2001, hereafter KW01) and the theoretical line separating purely star formation galaxies, Seyferts, LINERs and composite galaxies of Kewley et al. (2006, hereafter KW06 line). The empirical line dividing star-forming galaxies and composite objects of Kauffmann et al. (2003, hereafter KF03) was also included.

Since shocks could explain the LINER-like spectra (Dopita & Sutherland, 1995; Heckman, 1980; Dopita et al., 2015), we also added to the diagnostic diagrams shock models calculated with the mappings iii shock and photoionization code from Allen et al. (2008). These shock models include a range of chemical abundance sets, pre-shock densities from 0.01 to 1000 cm*-3*, velocities up to 1000 km/s, magnetic fields from 10*-10* to 10 -4 G, and magnetic parameters (B/$n^{1/2}$) from 10*-4* to 10 $\mu$G cm3/2. Since, the electron density ($n_{e}$) of the pre-shock material is an important physical quantity that affects the emission spectrum in radiative shocks, we estimated $n_{e}$ using [S II]$\lambda\lambda$ 6717$\AA$, 6731$\AA$ emission line ratios (Osterbrock & Ferland, 2006; Panuzzo et al., 2011) measured in the regions 1 to 4. We found 35 $<$ $n_{e}$ $<$200, the lower value corresponds to the region 4, while the higher value was found in the region 3, thus we adopted a mean value of $n_{e}$=100 cm*-3*. The models consistent with our observational data are those of shock without precursor, with solar metallicity and with pre-shock density of $n_{e}$=100 cm*-3*.

In Figure 8 we show the ratios obtained for all the spaxels along the FoV (yellow) and highlight those of the nuclear region (black). It is clear that the nuclear region presents some discrepancy with respect to the circumnuclear one, suggesting a mixing of ionization mechanisms, since the black circles are in the LINER region, while the circumnuclear (yellow) lies in the shock zone.

In addition, we investigate in detail the excitation mechanism of the selected regions 1, 2, 3 and 4 identified in the H$\alpha$ channel maps (Figure 5). For each one of these regions we constructed diagnostic diagrams involving the line ratios [O iii]/H$\beta~{}\times$ [S ii]/H$\alpha$ and [O iii]/H$\beta~{}\times$ [N ii]/H$\alpha$ (Figure 9). For the diagnostic diagrams using the [SII]/H$\alpha$ ratio (which is more sensitive to shocks), we found shock velocities $>$ 550 km s*-1*, while the shock models for [NII]/H$\alpha$ are not able to predict the observed values for regions 1, 3, and 4. However, region 2 can be explained by models with velocities of $\sim$400 $\rm km\,s^{-1}$, for the [NII]/H$\alpha$ ratio. The above results suggests that the dominant excitation mechanism in the identified regions is compatible with shock waves with velocity $>$ 550 $\rm km\,s^{-1}$(for the [SII]/H$\alpha$ ratio) moving in a environment with density of $n_{e}$ $=$ 100 cm*-3*. Therefore, in the inner kpc of this galaxy there are two ionization mechanisms taking place: in the center ($\sim$200 pc) the emission lines are dominated by photoionization due to a LLAGN, while in the circumnuclear region the gas emission is can be shock dominated. In addition this result is in agreement with the fact that we are finding a significant fraction of the FC in the central region (see Fig. 2) and with the broad component that we found for H$\alpha$ (see Figure. 6-Nuc), that indicates the presence of an AGN.

6 The gas velocity field of NGC 5044: determination of the supermassive black hole mass

In order to understand the gas kinematics, we have modelled the observed velocity field, assuming that the ionized gas follows circular orbits in a plane as proposed by Bertola et al. (1991), summarized by the following equation:

[TABLE]

This results in a velocity curve that increases linearly at small radii and becomes proportional to $r^{1-p}$ for the external regions. $R$ and $\psi$ are the radial and angular coordinates of a given pixel in the plane of the sky; $A$ is related to the amplitude of the rotation curve; $c_{0}$ is a concentration parameter. The parameter $p$ defines the form of the rotation curve, varying in the range between 1 (logarithm potential) and 1.5 (Keplerian potential), which is the range of values expected for galaxies (Bertola et al., 1991); The parameter Vs is the systemic velocity; $\psi_{0}$ is the position angle of the line of nodes; $\theta$ is the disk inclination ($\theta$ = 0 for face-on disks); finally, there are two implicit parameters: the coordinates of the kinematic center, $R_{cx}$ and $R_{cy}$. The equation 3 has a degeneration between $A$ and $\theta$ parameters, which makes not reliable the $\theta$ value derived to the fit. Then, we did an estimative for $\theta$ of 42$\degr\pm 2\degr$ from the outer isophotes of the H$\alpha$ flux image. Thus, this parameter was fixed for the fit.

We used a Levenberg-Marquardt method to fit the observed velocity field. The weighted fit was done considering the error map of $\delta_{v_{r}}$ (Fig. 10d). The fitted parameters and their respectively errors are given in Table  4. The $R_{cx}$ and $R_{cy}$ are not listed in this Table, instead of them, we give the difference between kinematic and photometric centres, in the sky plane, $\Delta\,x$ and $\Delta\,y$. The reduce $\chi^{2}$ of the fit is 10.8, this high value is due to the perturbations of the observed velocity field, and is reflected in the residual map (Fig. 10c), which presents non-rotational strong kinematic components, they are discussed and analysed in detail in Sec. 7.

The galaxy kinematic centre has a small offset, of 18 pc to NE direction, with respect to its photometric centre. Considering that our angular resolution is $\sim$ 125 pc this offset is negligible and we conclude that the kinematic centre coincides (within the uncertainties) with the photometric nucleus.

The rotating disk model that provided the best fit of the observed velocities is shown in Fig. 10b. The rotation curve derived from this model is presented in Figure 11. It is clear that the gas reaches its maximum rotational velocity ( 240 km s*-1*) at a distance of 136 pc. In addition, the derived value for the $p$ parameter (p$\sim$1.5) indicates that the gas distribution has a Keplerian behaviour and thus the dynamic in this region must be dominated by a high mass concentration. Assuming the Keplerian approximation for the gas motion ($M=r\times V_{{\rm rot}}^{2}/G$), we estimate a mass of $1.9\pm 0.9\times 10^{9}M_{\odot}$ within the inner 136 pc.

The value of this mass concentration can be verified using the M-$\sigma$ relation (Kormendy & Ho, 2013) as follows:

[TABLE]

where $\sigma_{\star}$ is the stellar velocity dispersion measured in an aperture with radius of $R_{e}$/8, with $R_{e}$ being the half-light radius. We derived this radius for NGC 5044 from its brightness surface profile, using a $K_{s}$ band image taken from 2MASS Extended Source Image Catalogue (Skrutskie et al., 2006). The obtained value is $R_{e}$=29″, which corresponds to 5.1 kpc, the required value of $R_{e}$/8 is $\sim$ 600 pc, that is close to the datacube FoV. The $\sigma_{\star}$ was determined from the absorption lines Mgb$\lambda$5167, FeI$\lambda$5270 and FeI$\lambda$5335 measured in a single integrated spectrum obtained collapsing all the non stellar subtracted cube spectra, since we are only interested to measure the FWHM of these stellar lines. The FWHM of the lines was corrected for instrumental broadening. The resulting $\sigma_{\star}$ is the mean weighted by the flux of each absorption line, and the uncertainties was estimated from the standard deviation of these measurements. Then, the found value of $\sigma_{\star}$ for NGC 5044 is $300\pm 57$ $\rm km\,s^{-1}$. This value is in agreement with the 250 $\rm km\,s^{-1}$found for the galaxy center in Caon et al. (2000). The $M_{bh}$ derived using this $\sigma_{\star}$ is $1.8\pm 1.6\times 10^{9}\,M_{\odot}$, which is in full agreement with the value obtained with the Keplerian approximation.

With the purpose of testing the possibility that the peculiarities in the velocity field obtained in Sec. 4 [e.g. a steep rise toward the center (300 - 600 $\rm km\,s^{-1}$), and the rotation curve reaching its maximum value at a small radius (136 pc)] are associated with the presence of a black hole, we plotted in Figure 11 the Keplerian rotation curve expected for a SMBH with M=1.8$\pm 1.6\times 10^{9}\,M_{\odot}$ (dash-dotted line), the observed (red line) and modelled (black line) rotation curves for the central region of the galaxy. We also include the rotation curve derived from the Sérsic profile (dashed line) using the K band brightness surface profile (see Hernandez et al., 2013, 2015, for details on methodology). From Figure 11, it is clear that for the inner 136 pc (dashed vertical line), the central kinematics is dominated by the SMBH potential. This would explain why the maximum rotation curve of the gas and the steep rise in the dispersion velocity profile is observed inside this small radius. Otherwise, without the presence of a SMBH the observed rotation curve would follows that one derived from the Sérsic profile. A similar kinematic behaviors as that we found for NGC 5044 have been observed in other early type galaxies, for example the elliptical M 87 that harbor a SMBH (3$\times 10^{9}\,M_{\odot}$ Macchetto et al., 1997; Ho & Kormendy, 2000) and the lenticular galaxy NGC 3115 ($10^{9}\,M_{\odot}$ Kormendy & Ho, 2000). Taking the evidences presented here, we conclude that the nuclear gas motions are driven by the gravitational potential of the central SMBH, with mass of $1.8\pm 1.6\times 10^{9}\,M_{\odot}$.

7 Evidence of gas inflow in NGC 5044

In order to derive the non-rotational velocities, resulting from radial or perpendicular motions, we used the residual velocity map obtained by subtracting the velocity field model from the observed one (Fig. 10c). We can distinguish in the residual map two blueshifted regions with velocities around $\sim$ $-40$ $\rm km\,s^{-1}$, one at east and other at west. Both regions show a peak of $-80$ $\rm km\,s^{-1}$. On the other hand, two redshifted structures are clearly observed along the galaxy’s line-of-nodes (north and south), with velocities of 40 $\rm km\,s^{-1}$. These regions have the velocity dispersion of 150 $\rm km\,s^{-1}$(see Figure10e). The detection of these structures is reliable, because their velocities are higher than the uncertainties ($\delta_{v_{r}}$) as seen in the velocity field uncertainty map included in Fig. 10 (panel d).

With the purpose of interpreting the structures described above, we used the Shape software (Steffen et al., 2010), which allows the user to create interactively 3D spatial structures with their respective velocity fields. The outcome is a projected synthetic velocity field image with a similar orientation to the observer. The output velocities are integrated along the line of sight. We presents in Figure 12 the 3D model for NGC 5044. We assume a rotating disk represented by a cylinder, whose height is one tenth of the radius and its velocity field is that obtained in Section 6, whereas we assume for all gas filamentary structures a geometry of cone with an aperture of $15\degr$, where the gas inside it has a velocity vector oriented toward the center. The orientations of the cones and the magnitudes of their respective velocity vectors were constrained requiring that projected velocities of these structure, along of the line of sight, match the residual map. We found that the best orientation for the cones is when: a) The north and south cone-filaments are inclined in 20$\degr$ with respect to the disk and they have a PA with respect to the line-of-node of 0$\degr$ and 180$\degr$, respectively, and the magnitudes of their respective velocity vectors are $80$ $\rm km\,s^{-1}$ b) the cone associated with the east filament is inclined 26$\degr$ with respect to the disk, has a PA of 90$\degr$, and the magnitude of its velocity vector is $105$ $\rm km\,s^{-1}$; c) the cone that represents the west filament is inclined in $-62\degr$ with respect to the disk, has a PA of 270$\degr$, and the magnitude of its velocity vector is $105$ $\rm km\,s^{-1}$. The resulting geometrical configuration, along the line of sight, to represent the gas filamentary structures is shown in Fig. 12a. The synthetic model of the velocity field is presented in in Fig. 12b. It is created from the sum of the disk and the filamentary structures models, which would represent the observed velocity field as seen in Fig. 10a. We can see that the model is able to recovery the asymmetry in the observed velocity field. In addition, we present in Fig. 12c the projected velocities of the cones along the line of sight. It reproduces well the values of residual velocity map (Fig. 10c). This suggest that a suitable interpretation of the residual map could be that the gas in the filamentary structures is falling toward the galaxy center. This result is in agreement with the findings of David et al. (2014), which suggested that some of the molecular CO clouds they detect are infalling towards the galaxy in a ballistic trajectory. However, we should be caution with our results since they are a simple model approach to reproduce the complex velocity field of the inner 600 pc of NGC5044.

Using the above results we are able to make an estimative of the inflow rate of the gas in the filamentary structure from their observed H$\alpha$ luminosity and the mean magnitude of the velocity of the cone-filaments. Thus, first we integrated the H$\alpha$ flux in all FoV and found ${F_{H\alpha}}$= 1.65$\times$10*-12* erg cm*-2* s*-1*, which at the galaxy distance (Table 1) results in ${L_{H\alpha}}$=2.7$\times$1041 erg s*-1*. Note that the value obtained here is higher than that found by Macchetto et al. (1996) integrated over the entire galaxy. We attribute this difference to indirect calibration approach used by Macchetto et al. (1996) .

The ionized gas mass, $M_{HII}$, can be estimated as follows:

[TABLE]

where, $m_{p}$ is the proton mass and $\alpha^{eff}_{H\alpha}$ is the effective recombination coefficient for H$\alpha$. Then, by assuming the “case B” for the recombination (Osterbrock & Ferland, 2006), adopting a Te=10000 K and the mean ne= 100 cm*-3* around of FoV, we found a ${M_{HII}}$=6.38$\times$10${}^{6}M_{\odot}$. Finally, the inflow rate is calculated with the following expression:

[TABLE]

where $N_{HII}$ is the density of ionized gas, $\upsilon$ the velocity of the inflowing material, $A$ is the area of the cone base, and the factor 4 is due to the number of cones. $N_{HII}$ was estimated adopting a volume of 0.2 kpc3, which was calculated from the equivalent radius of the physical area of the FoV, $r_{FoV}=0.41\,kpc^{2}$. $\upsilon=93$ km s*-1* is the mean value of all cones. $A=0.03\,kpc^{2}$ was calculated from the $r_{FoV}$ and aperture angle of the cones ($15\degr$). Then, we found a value of $\dot{M}_{inflow}\approx 0.4\,M_{\sun}\,yr^{-1}$. We can now compare it to the mass accretion rate necessary to power the active nucleus, by using

[TABLE]

where $L_{bol}$ is the bolometric luminosity, $\eta$ is the efficiency of conversion of the rest-mass energy of the accreted material into radiation, and $c$ is the light speed. We approximated the bolometric luminosity as $L_{bol}\approx 100L_{H\alpha}\approx 2.7\times 10^{43}$ erg s*-1* (e.g., Ho 1999, et al. 2001). We adopted a value of 0.1 for $\eta$, which is a typical value for a optically thick and geometrically thin accretion disc (e.g. Frank, King & Raine 2002). Then, we obtain an accretion rate of $\dot{M}\approx 0.005\,M_{\sun}\,yr^{-1}$. This value is a factor of $10^{2}$ lower than the inflow rate calculated.

One question which rises now is: are these filamentary structures an extension of the larger-scale (kpc) filaments earlier studied by Macchetto et al. (1996)? Temi et al. (2007) found a remarkably spatial correlation between the filamentary structures in H$\alpha$ and dust detected at 8 and 70 $\mu$m. These authors claimed that the dust present in the filaments most likely comes from the galaxy centre. Based on these results, they proposed that NGC 5044 has suffered a recently (107 yr ago) feedback where the dust located in the nuclear region was disrupted, heated and transported outward to several kiloparsecs. Evidences of bubbles were found by Gastaldello et al. (2009) which reported that the inner 10 kpc of this galaxy harbors a pair of cavities together with a set of bright X-ray filaments coincident with H$\alpha$ and dust emission. In addition, Gastaldello et al. (2009) make an estimate of the release time for the bubbles as 1.2$\times$107 yr, supporting the scenario proposed by Temi et al. (2007) for this galaxy. Besides, our results point to ionized gas flowing towards the galaxy center, which one can associate with the kpc scale filamentary structure reported for this galaxy. In addition, we found strong evidence for the presence of an accreting SMBH in the nuclear region of NGC 5044, besides a broad component for H$\alpha$ therefore, we speculate that this gas inflow would be originated and transported from large kpc scale to the nucleus of NGC 5044 through the filamentary structures and part of it being accreted by the SMBH and making NCG 5044 active. This would be in agreement with the scenario where the cooling flows into central galaxy of galaxy clusters (which is the case of NGC 5044) are regulated by cyclic feedback episodes (e.g. Peterson & Fabian, 2006, and references therein).

8 CONCLUSIONS

We present here, for the first time, high spatial resolution Gemini GMOS/IFU data of the elliptical galaxy NGC 5044 to map the stellar population, emission line fluxes distribution and gas kinematics in the inner kpc, with an spatial resolution of 125 pc. Our mains conclusions are:

• •

The continuum emission of the inner kpc of NGC 5044 is dominated by an old stellar population. We also detect an intermediate age population ($\sim$ 40% of an 900 Myr SSP) surrounding the galaxy nucleus. A featureless contribution of $\sim$ 20% was found in the the galaxy centre, consistent with the detection of a broad component in H$\alpha$. Besides, the maps of mean age and mean Z show a trend consistent with the stellar population becoming more metal poor and younger with increasing distance from the galaxy center.

• •

The analysis of the emission line intensities revels that in the inner kpc of the NGC 5044, the gas distribution is irregular and the H$\alpha$ emission follows the extended filaments. The [NII] and [SII] emission is concentrated in the nuclear region, while the [OIII] emission is detected over the whole FoV.

• •

To fit the central region H$\alpha$ emission lines a broad component is required, with FWHM $\sim$ 3000 $\rm km\,s^{-1}$  confirming that the featureless continuum emission detected in the stellar population fits indeed is due to a low luminosity active galactic nuclei.

• •

The FWHM of H$\alpha$ narrow component present strong variations over the FoV, from FWHM $\sim$ 150 $\rm km\,s^{-1}$ at the outer regions to 1080 $\rm km\,s^{-1}$ in the galaxy centre.

• •

The dominant excitation mechanism in the filaments is compatible with shock waves with velocity 550 $\rm km\,s^{-1}$  while in the centre it is explained by a LLAGN. Furthermore, we found that the brighter peak located at east (region 1) and at the south (region 3) in the channel maps are co-spatial with the CO emission detected by ALMA.

• •

From the H$\alpha$ velocity field map, we found evidence of rotating gas. Assuming Keplerian rotation we estimated a mass of 1.9$\times 10^{9}M_{\odot}$ within 136 pc. Beside, using $\sigma_{\star}$ of 279 $\rm km\,s^{-1}$and M-$\sigma$ relation the estimated $M_{bh}$ found is $1.8\pm 1.6\times 10^{9}\,M_{\odot}$ .

• •

Filamentary structures, with non-circular motions around $\mid$40$\mid$ $\rm km\,s^{-1}$are revelled in the residual velocity field map, obtained subtracting the disk model from the observed radial velocity field. By using 3D model for these structures we conclude that the gas inside them is falling towards the nucleus. The inflow rate of 0.4 $M_{\odot}\,yr^{-1}$, was derived from H$\alpha$ luminosity of $2.7\times 10^{41}$ erg/s and with the mean inflow velocity, $\sim$90 km/s, estimated for the 3D model. We point out that the inflow rate is $10^{2}$ larger than the accretion rate necessary to power the LLAGN.

Acknowledgements

We thank the anonymous referee for his/her careful reading of our manuscript and for his/her insightful comments and suggestions that helped to improve the manuscript. J. A. H.J. thanks to Brazilian institution CNPq for financial support through postdoctoral fellowships (projects 158962/2014-1 and 150237/2017-0). R.R. Thanks to CNPq and FAPERGS for partial financial support to this project. R. A. R. thanks financial support from Brazilian institution CNPq (project 303373/2016-4). T.V.R. acknowledges CNPq for the financial support under the grant 304321/2016-8. Based on observations obtained at the Gemini Observatory, which is operated by the Association of Universities for Research in Astronomy, Inc., under a cooperative agreement with the NSF on behalf of the Gemini partnership: the National Science Foundation (United States), the National Research Council (Canada), CONICYT (Chile), Ministerio de Ciencia, Tecnología e Innovación Productiva (Argentina), and Ministério da Ciência, Tecnologia e Inovação (Brazil).