Cepheid Metallicity in the Leavitt Law (C- MetaLL) survey – II. High-resolution spectroscopy of the most metal poor Galactic Cepheids Free

ABSTRACT

1 INTRODUCTION

Classical Cepheids (DCEPs) are fundamental astrophysical objects. Since their discovery, the period–luminosity (PL) and period–Wesenheit¹ (PW) relations that hold for these pulsators, represent the base for the cosmic distance scale (e.g. Leavitt & Pickering 1912; Madore 1982; Caputo et al. 2000; Riess et al. 2016). Once calibrated by means of geometric methods such as trigonometric parallaxes, eclipsing binaries and water masers, these relations can be used to calibrate secondary distance indicators like Type Ia Supernovae (SNe), which are sufficiently powerful to reach the unperturbed Hubble flow, allowing us to measure the Hubble constant (H₀, see eg. Sandage & Tammann 2006; Freedman et al. 2012; Riess et al. 2016, 2019, 2021a, and references therein).

In recent years this topic has been at the centre of a heated debate as the values of H₀ based on the cosmic distance scale (e.g. Verde, Treu & Riess 2019; Riess et al. 2021b, and references therein) are in significant disagreement with those estimated from the Planck cosmic microwave background (CMB) measurements under the flat Λ cold dark matter (ΛCDM) model. The latest estimate of H₀ from the cosmic distance scale is H₀ = 73.04 ± 1.04 km s⁻¹ Mpc⁻¹ (Riess et al. 2021a), which is 5σ different from the value estimated by Planck + ΛCMB, namely H₀ = 67.4 ± 0.5 km s⁻¹ Mpc⁻¹ (Planck Collaboration 2020). This discrepancy, known as the H₀ tension, has still unknown origins, in spite of many observational and theoretical efforts to study the possible causes as well as residual systematics(see e.g. Dainotti et al. 2021; Freedman 2021; Riess et al. 2021b, and reference therein).

In this context, it is crucial to identify any residual systematic effect which could in principle affect the cosmic distance scale path to the measure of H₀. One of the most controversial candidates is the metallicity dependence of the DCEP PL and PW relations. Several recent results based on Gaia mission Data Release 2 (DR2 Gaia Collaboration 2016, 2018) and Early Data Release 3 (EDR3 Gaia Collaboration 2021) parallaxes are indeed in disagreement one each other (see e.g. Groenewegen 2018; Ripepi et al. 2019, 2020; Breuval et al. 2021; Riess et al. 2021a; Ripepi et al. 2021a, 2022). To overcome these difficulties and to establish accurately the metallicity dependence of the DCEPs PL and PW relations and its impact on the measure of H₀, we undertook a project dubbed C-MetaLL (Cepheid – Metallicity in the Leavitt Law) which is fully described in Ripepi et al. (2021a). One of the immediate objectives of the project is to measure the chemical abundance of a sample of 250–300 Galactic DCEPs through high-resolution spectroscopy, specifically aiming at enlarging as much as possible the metallicity range of the targets towards the metal-poor regime ([Fe/H] < −0.4 dex), where only a few objects are present in the literature.

Here we present the spectroscopic observations of a considerable sample (i.e. 65 objects) of recently discovered faint DCEPs whose Galactocentric distance, R_GC > 15 kpc, makes them excellent metal-poor candidates (see Section 2.1 for details).

The DCEP sample discussed in this work has not only relevance for the cosmic distance scale, but also for Galactic studies. Indeed, the accurate distances which can be measured for DCEPs allow one to carry out the chemical tagging of the disc and in particular to study how the abundance of the different chemical elements varies with R_GC, from the central regions of the Milky Way (MW) towards the outer regions of the disc. Many studies exist already in the literature facing this problem with a variety of tracers other than DCEPs, such as open clusters (OCs) (e.g. Cunha et al. 2016; Netopil et al. 2016; Casamiquela et al. 2019; Donor et al. 2020; Spina et al. 2021, just to mention the most recent ones). Focusing on the most recent works using DCEPs, we notice that their majority (Luck & Lambert 2011; Genovali et al. 2014; Lemasle et al. 2018; Luck 2018; da Silva et al. 2022; Ripepi et al. 2022) agree on measured iron gradients in the interval 0.05–0.06 dex kpc⁻¹. However, all these investigations followed the radial gradient of the MW disc only up to about R_GC ∼ 13–14 kpc, not exploring the outer and possibly most metal-poor regions of the disc, with the exception of Minniti et al. (2020), who obtained a metallicity estimate for three DCEPs with 15 < R_GC < 22 kpc, beyond the Galactic bulge. Our sample will allow us increase significantly the number of DCEPs with R_GC >15 kpc and get new insights on the metallicity gradient of the MW disc. The application of our sample to the determination of the metallicity dependence of the DCEPs PL/PW relations will be the subject of a separate paper.

2 DATA

2.1 Selection of the targets

Our targets have different characteristics due to the displacement of the MW disc during the different seasons. During the southern winter we can observe in the direction of the Galactic centre, while the opposite is true for the summer season. Given the Galactic gradient, the southern winter and summer seasons allow us to observe DCEPs with solar or super-solar and sub-solar metallicity, respectively. The targets towards the Galactic centre were selected among the list published by Skowron et al. (2019), requiring that the relative error on the parallax published in the Gaia DR2 were better than 20 per cent and that G < 13 mag (note that EDR3 was not available at the time of proposal submission). This limit ensures that the target were observable in less than 1 h (the maximum allowed duration of an ‘Observing Block’ at ESO) in any weather condition (so called ESO ‘filler programme’). To choose the winter sample, we used again the Skowron et al. (2019) catalogue, and by relying on the distances therein, we selected the targets with Galactocentric distance larger than 15 kpc. In this way, thanks to the metallicity gradient of the Galactic disc (see e.g. Luck & Lambert 2011; Kovtyukh & Gorlova 2000; Ripepi et al. 2022, and references therein), we could reasonably expect that the large majority of these DCEPs had [Fe/H] values lower than −0.4 dex, which is one of our main aims.

2.2 Observations and data reduction

The spectra analysed in this work have been obtained in the context of two ESO (European Southern Observatory) proposals, namely P105.20MX.001 and P106.2129.001. The observations foreseen for P105 were postponed due to pandemic issues and were collected in service mode between 2021 January 2 and September 13. For P106 the observations were carried out in virtual visitor mode in the period 2020 December 12–19 under very good sky conditions. We used the UVES (Ultraviolet and Visual Echelle Spectrograph)² instrument, attached at the UT2 (Unit Telescope 2) of VLT (Very Large Telescope), placed at Paranal (Chile). We used the red arm equipped with the grism CD#3, covering the wavelength interval 4760–6840 Å, and the central wavelength at 5800 Å. We adopted the 1 arcsec slit, which provides a dispersion of R∼47 000. The exposure times and the signal-to-noise (S/N) for each target are listed in Table 1, together with the identification of the targets, coordinates, period, G magnitude, and mode of pulsation.

The data reduction was carried out using the standard UVES pipeline version 6.1.6. which provides fully calibrated spectra. The spectra were normalized using the iraftask continuum after dividing each spectrum into intervals (usually 3/4) in order to have a better continuum detection in each interval.

3 ABUNDANCE ANALYSIS

3.1 Stellar parameters

The first step when measuring chemical abundances is the estimation of the main atmospheric parameters, namely the effective temperature (T_eff), surface gravity (log g), microturbulent velocity (ξ), and line broadening parameter (v_br), i.e. the combined effects of macroturbulence and rotational velocity (in the case of DCEPs, macroturbulence often represents the dominating factor).

A useful tool extensively adopted in the literature to evaluate the effective temperature, is the line depth ratios (LDR) method (Kovtyukh & Gorlova 2000). This method has the advantage of being sensitive to temperature variations, but not to abundances and interstellar reddening. Typically in our targets we measured about 32 LDRs for each spectrum.

For some targets, because of their low metallicity, the number of couples of spectral lines useful for the LDR method is not enough to guarantee accurate results. For these stars the effective temperature can be determined by imposing excitation equilibrium, that is, by imposing that there is no residual correlation between the iron abundance and the excitation potential of the neutral iron lines (see e.g. Mucciarelli & Bonifacio 2020). To check the consistency of both methods, we computed the effective temperatures by using excitation potentials for all the targets for which LDRs gave accurate results. In total we selected 55 targets for which we calculated T_eff with both methods. The average difference among all these values is ΔT =  17 ±  160 K, thus we can conclude that both methods are consistent. The results of this comparison are shown in Fig. 1.

Figure 1.

Effective temperatures derived by excitation potential of neutral iron lines versus those derived by LDR. The upper panel shows the direct comparison between the two estimates, while the lower panel displays their differences.

Open in new tabDownload slide

For the other parameters (ξ and log g) an iterative approach was adopted: microturbulences were estimated by demanding the iron abundances do not depend on the equivalent widths (EWs), that is, the slope of the [Fe/H] as a function of EWs is null. For this purpose, we first measured the equivalent widths of a sample of 145 Fe i lines using a python3 semi-automatic custom routine. The lines sample was extracted from the line list published by Romaniello et al. (2008) and the routine minimizes errors in the continuum estimation on the wings of the spectral lines. The conversion of the EWs in abundances has been performed through the width9 code (Kurucz & Avrett 1981) applied to the corresponding atmospheric model calculated by using atlas9 (Kurucz 1993). In this calculation we did not consider the influence of log  g since neutral iron lines are insensitive to it. Then, the surface gravities were estimated with a similar iterative procedure imposing the ionization equilibrium between Fe i and Fe ii. The adopted list of 24 Fe ii lines was extracted from Romaniello et al. (2008).

The atmospheric parameters estimated, and summarized in Table 2, were used as input values for the abundance analysis presented in the next section.

3.2 Abundances

In order to avoid problems from spectral line blending caused by line broadening, a spectral synthesis technique was applied to our spectra. Synthetic spectra were generated in three steps: (i) plane parallel local thermodynamic equilibrium atmosphere models were computed using the ATLAS9 code Kurucz (1993), using the stellar parameters in Table 2; (ii) stellar spectra were synthesized by using SYNTHE Kurucz & Avrett (1981); (iii) the synthetic spectra were convoluted for instrumental and line broadening. This was evaluated by matching the synthetic line profiles to a selected set of the observed metal lines.

For a total of 24 different chemical elements it was possible to detect spectral lines used for the estimation of the abundances. For all elements we performed the following analysis: we divided the observed spectra into intervals, 25 Å or 50 Å wide, and derived the abundances in each interval by performing a χ² minimization of the differences between the observed and synthetic spectra. The minimization algorithm was written in IDL³ language, using the amoeba routine.

We considered several sources of uncertainties in our abundances. First, we evaluated the expected errors caused by variations in the fundamental stellar parameters of δT_eff = ±150, δlog g= ± 0.2 dex, and δξ ± 0.3 km s⁻¹. According to our simulations, those errors contribute ≈± 0.1 dex to the total error budget. Total errors were evaluated by summing in quadrature this value to the standard deviations obtained from the average abundances.

The adopted lists of spectral lines and atomic parameters are from Castelli & Hubrig (2004), who have updated the original parameters of Kurucz (1995). When necessary we also checked the NIST database (Ralchenko & Reader 2019).

An example of four spectra in the range λλ 6100–6150 Å is displayed in Fig. 2. With the aim of comparing the differences in the spectral line depth, each spectrum reports, as a side label, with the metallicity derived in our analysis (see Section 3.2).

Figure 2.

Example of high-resolution spectra in the range λλ 6100–6150 Å. Stars are ordered from top to bottom with decreasing metallicity (see Section 3.2 for its estimation). The vertical dashed lines identify the position of atomic lines, the meaning of the colours are: red for lines of vanadium, iron, and nickel, blue for lines of silicon and calcium, and black for the line of barium.

Open in new tabDownload slide

In Fig. 3 we show the distribution of the elements derived in this analysis in the form of histograms, where bins have been fixed to 0.15 dex, in order to be representative of the errors. The literature sample is also shown for comparison (see Section 4). For each element, the Gaussian fit has been over-imposed with the respective mean value and its full width at half-maximum (FWHM) is reported in the upper right corner of each box.

• Carbon, oxygen: The following spectral lines were used for carbon: λλ 4932.049, 5052.144, and 5380.0325 Å. The Gaussian fit reports a mean abundance of 0.80 dex below the solar one, with an asymmetrical tail towards higher values. For oxygen, the forbidden line O i at 6300.304 Å and the O i triplet at 6155-8 Å, obtaining a broader distribution centred at −0.37 dex. No nitrogen lines could be detected in our observed spectral range.

• Sodium: Six neutral lines have been found for this element: λλ 5682.647, 5688.217, 5889.930, 5895.910, 6154.230, and 6160.753 Å. The mean value is slightly under abundant, at −0.15 dex.

• Aluminum: Two aluminum neutral lines have been detected for our spectra, λλ 6696.018 and 6698.667 Å, obtaining a distribution slightly under abundant.

• α-elements: Magnesium (four Mg i spectral lines: λλ 5167.320, 5172.680, 5183.600, and 5528.405 Å), Silicon (four Si ii lines at λλ 5041.024, 5055.984, 6347.11, and 6371.370 Å), and Calcium (via Ca i lines, λλ 5598.480, 6162.170, 6166.433, 6169.042, 6169.583, 6462.567, 6471.662, 6493.781, and 6499.65 Å) result under-abundant, while Sulfur (three S ii at λλ 6743.585, 6748.153, and 6757.153 Å) and Titanium (several Ti i and Ti ii all over the spectrum) are consistent with about −0.30 dex under solar abundances although the latter has a wider distribution.

• Scandium: Five Sc ii lines were measured at λλ 5031.021, 5237.813, 5526.818, 5684.000, and 6245.621 Å. The resulting distribution is centred on about solar value.

• Iron peak elements: V, Cr, Mn, Fe, Co, Ni, Co and Zn: four neutral vanadium lines (λλ 5698.482, 5703.569, 6090.194, and 6243.088 Å) distributed around −0.13 dex. Whilst Cobalt is consistent with solar abundances (measured using three Co i lines at λλ 5266.490, 5352.000, and 6450,067 Å), manganese has a spread distribution around −0.66 dex (five Mn i lines at λλ 4823.515, 5341.057, 6013.458, 6016.638, and 6021.790 Å) and Chromium is slightly under abundant (with five Cr ii spectral lines: λλ 4824.131, 4876.339, 5204.506, 5206.037, and 5208.419 Å), Iron and Nickel (Ni i: λλ 5080.533, 5081.110, 5084.011, 5094.411, and 5099.931 Å) are significantly under abundant, with distributions centred at −0.53 and −0, 47 dex, respectively. We used two Cu i neutral lines at λλ 5105.500 and 5218.201 Å, with abundances consistent with the solar ones. For zinc only one neutral line could be measured at λ 6362.338 Å. The mean value of the distribution is centred on −0.86 dex.

• Yttrium and Zirconium: with four Y ii lines (λλ 4883.682, 4900.120, 5087.418, and 5402.774 Å) and one Zr ii line (λ 6114.853 Å) we obtained two distributions centred at −0.15 and 0.22 dex, respectively.

• Barium: As expected, using four Ba ii spectral lines at λλ 4934.100, 5853.625, 6141.713, and 6496.897 Å, we estimated abundances centred on −0.10 dex but distributed over a wide range, from ≈−1.5 to ≈1.0 dex.

• Rare earth elements: We obtained distributions centred at −0.09, −0.37, and −0.22 dex, respectively for Lanthanum (La ii lines at λλ 4921.776, 5114.559, and 5290.818 Å), Praseodymium (one Pr ii line at λ 5219.045 Å), and Neodymium (five Nd ii lines at λλ 4959.119, 4989.950, 5092.794, 5293.163, and 5688.518 Å).

Figure 3.

Histograms of the distribution of the chemical abundances derived in this study (blue histograms) compared with literature data as defined in Section 4.1 (grey histograms). Gaussian fits have been overimposed in our data (red dashed curves) and the respective mean value and FWHM reported in each box.

Open in new tabDownload slide

A comprehensive list of all estimated abundances is shown in Table A1.

The differences between our results and the literature values are largely due to the location of a significant fraction of our targets in the Galactic anticentre region, characterized by significantly lower abundances than in regions closer to the Galactic centre (see discussion in Section 4).

4 COMPARISON WITH THE LITERATURE

4.1 Literature sample

For the following analysis, we complemented the sample studied in this work with stars from the literature. More in detail, we considered the large compilation of homogenized literature iron abundances for 436 DCEPs presented by Groenewegen (2018, G18 hereinafter), complemented with literature results for a few additional stars by Gaia Collaboration (2017, GC17 hereinafter). To these data we added the sample of 49 stars presented in our previous works (collectively called R21 hereinafter Catanzaro et al. 2020; Ripepi et al. 2021a, b) and the recently published large sample of 104 stars by Kovtyukh et al. (2022, K22 hereinafter). There is almost no overlap between the G18/GC17 and the R21 sample, apart from the stars V5567 Sgr and X Sct (see Ripepi et al. 2021a). On the contrary, there is some overlap between K22 and the other samples, including the data set presented in this paper. More precisely, K22 has 13, 2, 14, and 2 stars in common with G18, GC17, R21, and this paper, respectively. In order to use all the data together, we seek for any systematic difference between the K22 and the other data sets. Albeit there is a large scatter, it can be seen that K22 on average tends to overestimate the [Fe/H] value by 0.06 ± 0.02 dex (the error is the standard deviation of the mean). In order to obtain a literature homogenized sample, we decided to remove from the K22 sample the stars in common with the other data sets and to add −0.06 dex to the [Fe/H] of the remaining 73 stars. Therefore, the total sample of DCEPs with [Fe/H] from high-resolution spectroscopy is therefore composed of 637 objects.

Concerning the elements other than iron that are not present in G18 or GC17, we adopted the homogeneous and extensive list published by Luck (2018, L18 hereinafter) for 435 Galactic DCEPs and merged it with that of K22, as performed above for iron. There were nine stars in common between L18 and K22 which were removed from the merged list.

4.2 Element by element comparison

In Fig. 4 we plot both our abundances and those from the literature, expressed in the form [X/Fe], as a function of metallicity, expressed as [Fe/H]. It is worth mentioning that, as can be seen also in Fig. 3, most of our stars cover the metallicity region between −0.8/−0.4 dex, previously less populated.

Figure 4.

Chemical element abundances (expressed as [X/Fe]) as a function of the iron content for both literature and our sample of stars. Red and grey points are the observed and literature abundances, respectively. The blue dotted horizontal lines point out the solar reference. The corresponding element of each plot is written on the top right.

Open in new tabDownload slide

In the following, we describe the behavior of each element with respect to the iron content, dividing each group of elements in a similar way as listed in Section 3.2.

• Carbon results from the first dredge-up Iben Jr (1967), where the incomplete CN-process outcomes mix to the surface. As for oxygen, it is mostly released by type II supernovae (SNe II), whose progenitors are young massive stars. Due to their short lifetime, they pollute the interstellar medium essentially instantaneously with respect to the timescale of Galactic evolution, so we expect higher abundances for lower metallicities, with decreasing values as the metallicity increases and other slower mechanisms start to have a major impact. What we find is a negative trend for oxygen, with metal-poor stars having over solar abundances, and an almost flat distribution for carbon.

• Both sodium and aluminum are odd-z elements that should be produced by core-collapse SNe. Nevertheless, we find a relatively flat trend for sodium and an average overabundance, interpreted by Sasselov (1986) as the result of the Na-Ne cycle for luminous stars. On the other hand, we find for Al an almost flat trend down to [Fe/H] ≈ −0.5 and a rise of abundance toward lower metallicities.

• α-elements (Mg, Si, S, Ca, Ti): As for oxygen, the α-elements production is due to core collapse supernovae, so we should expect a similar trend as oxygen. However, in Fig. 5 we plotted α-elements versus metallicity and we observe a general increase of their abundance toward lower metallicities.

• Scandium can be seen as an intermediate element between the α elements and the iron peak group. Similar to α elements, Scandium is produced in the innermost ejected layers of core-collapse SNe (type II) as reviewed by Romano et al. (2010), while the contribution from type Ia SNe seems to be negligible Clayton (2003). For this element we find a quite flat distribution over the metallicities, with the exception of the most metal poor of our sample, that seem to have an overabundance of scandium.

• Iron peak elements (V, Cr, Mn, Co, Ni, Cu, Zn) are those elements close to Iron, produced especially by Type Ia supernovae (SN Ia), the final stage of white dwarfs that undergo a binary merging. In general we observe a increase of abundances toward lower metallicities for elements like vanadium, chromium, cobalt and nickel, an almost flat trend for elements such as manganese and copper, while only zinc seems to increase with metallicities.

• For heavier elements, starting from Y, there are two main processes both based on neutron-capture. The so-called rapid neutron-process (r-process) whose possible channels refer to neutron star mergers, supernova explosions, and electron capture SN (see Argast et al. 2004; Surman et al. 2008; Korobkin et al. 2012; Cowan et al. 2021) and are responsible for the production of Y, Zr, La, Pr, and Nd, for which we have found negative trend with metallicities. The slow neutron-process (or s-process) which occurs in low and intermediate mass stars during the AGB phase (Karakas & Lugaro (2016)), is the primary source for barium formation. The trend of barium with respect to iron seems to have a maximum close to [Fe/H] ≈ −0.2 dex and decrease both toward higher and lower metallicities, but given the high scatter of the measurements no conclusions can be drawn. The cause of this high scatter of the measurements has been noticed in previous papers (see e.g. Simmerer et al. 2004; Andrievsky et al. 2013) and ascribed to the variation of microturbulent velocities from star to star.

Figure 5.

As in Fig. 4 but for α-elements abundance.

Open in new tabDownload slide

The behaviour of each element with respect to the iron abundance is in general agreement with recent results based on high-resolution spectroscopy of static stars belonging to the Galactic disc (see for example Hayden et al. 2017; Silva Aguirre et al. 2018; Spitoni et al. 2019, and references therein). A detailed discussion of the evolution of the chemical elements is beyond the scope of this paper. Here we only mention that, in general, the time-delay model (see Matteucci 2021, for a review) represents a robust interpretation for the observed abundance ratio versus metallicity ([Fe/H]) diagrams.

5 THE GALACTIC RADIAL GRADIENT

5.1 Distance estimate

5.2 Metal radial gradient of the MW disc

The relation between the iron abundance ([Fe/H]) and the Galactocentric radius (R_GC), is displayed in Fig. 6, for both our and literature samples. Our data extends significantly the range in R_GC over which it is now possible to investigate the metallicity disc gradient, reaching up to 18–20 kpc. It is noticeable the presence of the star ASAS J062939-1840.5 with [Fe/H] = −1.10 ± 0.19 dex, which is perfectly placed at the ideal extension of the metallicity disc gradient at 25 kpc.

Figure 6.

Galactic iron abundance radial gradient. Left: Grey and red dots show the data from the literature and this work, respectively. The big black dots with relative error-bars represent a binning of the data with bins of 1.33 kpc each. The light blue line shows a linear fitting to the data over the entire range of Galactocentric distances, while the two orange lines represent the fit to the data carried out by dividing the sample into two pieces with a break at R_GC = 9.25 kpc. The sun is shown with a yellow-black symbol. Right: Same as left, where all the data are in grey dots, and the lines represent the fit on the binning points.

Open in new tabDownload slide

We carried out a linear regression adopting the python ltsfit package (Cappellari et al. 2013). This software allows one to use weights on both axes and implement a robust outlier removal. We adopted a conservative 3σ clipping procedure, obtaining the [Fe/H]–R_GC relation. At the first glance, it appears that the fitting line does not represent well the data at low values of [Fe/H]. This could be due to the non-uniform sampling of the data. For this reason we operated a binning division of the entire sample in 12 intervals of around 1.33 kpc and subsequently estimated the coefficients of the [Fe/H]–R_GC relation. In the right-hand panel of Fig. 6 we highlight the binned points and the relative fitting line. Furthermore, different regimes might be identified in this diagram. Following the suggestion made by Genovali et al. (2014), it could be possible to distinguish an inner and an outer sample, with a break at about 9.25 kpc. The extension at lower metallicities presented in this work allows us to better verify this early proposal. Therefore, we decided to quantitatively estimate the metallicity radial gradient in an alternative way, that is dividing the sample into two sub-samples separated at R_GC = 9.25 kpc. We applied this division to both the data and the binning points. All the coefficients are listed at the beginning of Table 3. For each relation we estimated the root means square (rms) as a test of the goodness of the fit. The first three relations are over-imposed on the data on the left-hand panel in Fig. 6, the last three, corresponding to the fitted relation on the binned points, are plotted on the right-hand panel of the same figure. We point out that in both cases the inner and outer slopes differ at more than 1σ. The two-samples fits appear to reproduce well both the data and the binned point distributions. This occurrence suggests that the break in the Galactic disc metallicity gradient is plausible, but since the rms are comparable with that of the single line fit, for both the data and the binned points, we can conclude that a single line fit can certainly reproduce the data, but the hypothesis of the break cannot be ruled out on the basis of the present data sample.