About Metallicity Variations in the Local Galactic Interstellar Medium

The determination of the spatial variations of metallicity in the interstellar medium (ISM) across the Galactic disk is a fundamental constraint for understanding the importance of mixing processes, abundance gradients, and the chemical evolution of the Galaxy (i.e., Edmunds 1975; Roy & Kunth 1995). The presence of significant metallicity variations in the local Galactic ISM has been a matter of debate. Pioneer theoretical works devoted to estimating the timescale and efficiency of the mixing processes of heavy elements—and O in particular—in the ISM of galactic disks or in the Milky Way in particular, predicted very small variations (i.e., Edmunds 1975; Roy & Kunth 1995), several times smaller than those reported in early observational works. For example, studies on the radial gradients of the O abundance in the Galactic disk as those of Shaver et al. (1983) or Fich & Silkey (1991) were consistent with variations up to a factor two (0.3 dex) at spatial scales of ∼1 kpc. Later studies of the Galactic abundance gradient based on radio or far-infrared observations found a scatter of the O/H ratio around the gradient fit of about 0.16–0.20 dex or even higher (e.g., Quireza et al. 2006; Rudolph et al. 2006). Large variations were also reported in early works on stellar abundances in B-type stars as that by Rolleston et al. (1994), who found variations of about 0.7 dex in stars in clusters separated by distances of about 1 kpc. However, over time, thanks to the improvement in the depth and quality of observations and spectrum analysis techniques, the dispersion in the distribution of chemical abundances and metallicities have decreased considerably, indicating that much of this scatter—which was commonly interpreted as intrinsic—was actually largely due to observational and methodological errors. In the last decade, there has been growing evidence that H ii regions (e.g., Esteban & García-Rojas 2018; Arellano-Córdova et al. 2020, 2021) and Population I objects (O and B-type stars, classical Cepheids, or young clusters, i.e., Simón-Díaz 2010; Nieva & Przybilla 2012; Genovali et al. 2014; Luck 2018; Bragança et al. 2019; Donor et al. 2020) show fairly smaller variations in their chemical abundance distribution, especially once corrected for the radial abundance gradient.

However, there is no lack of evidence to the contrary. Very recently, De Cia et al. (2021) reported the existence of metallicity variations of up to a factor ten in the vicinity of the Sun, based on the study of neutral clouds along the line of sight of 25 O- or B-type stars located in or very close to the Galactic plane and within 3 kpc. They measure an average metallicity of −0.26 dex with respect to the solar one (55 per cent solar) finding values ranging between −0.76 dex and +0.26 dex, where ∼3/4 of the lines of sight show subsolar metallicities. De Cia et al. (2021) interpret the finding of low-metallicity variations as the product of the falling of high-velocity clouds of pristine gas onto the Galactic disk. If real, the metallicity distribution they observe would imply that the low-metallicity accreting gas does not efficiently mix into the ISM.

In this paper, we present and discuss a compilation of high-quality metallicity determinations for H ii regions and other young population objects to discuss their metallicity distribution and constrain the amplitude of possible spatial variations.

In Section 2 we present results on O abundance variations determined from the analysis of the spectra of Galactic and extragalactic H ii regions. In Section 3 we present and discuss recent results on metallicity variations in neutral clouds of the Galactic thin disk and the Magellanic Clouds and compare them with results for H ii regions. In Section 4 we compare the results for H ii regions and neutral clouds with metallicity variations determined from data for different local Galactic Population I objects. In Section 5, we discuss the likelihood of the existence of a large population of low-metallicity clouds in the local Galactic thin disk. Finally, in Section 6, we summarize our main conclusions.

2.1. Galactic H ii Regions

The most recent and accurate determination of the radial abundance gradients of the ionized gas phase of the ISM in the Milky Way is by Arellano-Córdova et al. (2020, 2021). Due to the characteristics of the sample and the unprecedented quality of their observations and analysis methodology, the results of that study are a very significant improvement over previous studies. Arellano-Córdova et al. (2020, 2021) use spectra of 42 H ii regions located at Galactocentric distances (R_G) from 4 to 17 kpc. Among the advantages of that work we can highlight: (a) the use of very deep spectra taken with 8–10 m telescopes; (b) the application of the same methodology and updated atomic data for the analysis of a large number of spectra; (c) the direct determination of electron temperature (T_e) for all objects; (d) the use of revised distances based on Gaia Data Release 2 (DR2) parallaxes (Bailer-Jones et al. 2018; Gaia Collaboration et al. 2018) of the stars associated with the nebulae, and (e) a careful selection of the most appropriate ionization correction factor (ICF) scheme for each chemical element, propagating the uncertainty of this magnitude in the determination of total abundances. Arellano-Córdova et al. (2020, 2021) estimate that the mean difference between the individual O abundances and that obtained from the linear fit of the radial O abundance gradient for the corresponding R_G of each object is 0.07 dex, a value consistent with the typical uncertainties of the determination of the O/H ratio in individual objects (the median is 0.08 dex). Considering this result, Arellano-Córdova et al. (2020, 2021) conclude that O is fairly well mixed in the Galactic ISM, discarding the presence of significant metallicity variations, at least in the quadrant of the Milky Way covered by their sample of H ii regions.

Figure 1 illustrates the improvement of the quality of the radial oxygen abundance gradient determined by Arellano-Córdova et al. (2021; black circles and continuous line) with respect to that derived by Rudolph et al. (2006; blue diamonds and the blue dashed line), one of the most widely used references of the Galactic O abundance gradient. Rudolph et al. (2006) use a compilation of abundances determined from optical spectra where T_e is obtained using different methods: the same optical spectrum, from radio observations (not cospatial, usually corresponding to the whole nebulae) or estimated indirectly by other means. The colored bands centered on the gradient lines represent the dispersion of the individual abundances around the gradient fit in each case; the gray area corresponds to the data by Arellano-Córdova et al. (2021), and the blue one to those by Rudolph et al. (2006). The dispersion obtained from the data by Rudolph et al. (2006) is about 0.24 dex, four times larger than the dispersion reported by Arellano-Córdova et al. (2021).

Zoom In Zoom Out Reset image size

2.2. Extragalactic H ii Regions

As it is said in Section 1, De Cia et al. (2021) reported the existence of large metallicity variations in the solar vicinity. Those authors estimate the metallicity of neutral clouds located along the line of sight of several O- or B-type stars located in the Galactic plane within 3 kpc of the Sun. In Figure 2, we show the spatial distribution around the Sun and onto the Galactic plane of the lines of sight used by De Cia et al. (2021) and the H ii regions observed by Arellano-Córdova et al. (2020, 2021) located at heliocentric distances <3 kpc. The figure includes 25 stars and 20 H ii regions. The heliocentric distances of the stars used by De Cia et al. (2021) have been taken from the Gaia DR2. It must be taken into account that the positions of the stars correspond to the upper limits of the true distance to the neutral clouds whose metallicity De Cia et al. (2021) estimate. The heliocentric distances to the H ii regions have been taken from Méndez-Delgado et al. (2022) and are derived from the Gaia Early Data Release 3 (hereinafter EDR3; Gaia Collaboration et al. 2021) parallaxes applying the probabilistic approach of Bailer-Jones et al. (2021).

Zoom In Zoom Out Reset image size

In Figure 3, we compare the metallicity of the objects represented in Figure 2 as a function of their corresponding R_G. We represent the relative metallicity of the objects with respect to the solar one, in the form: [M/H] = ${\mathrm{log}}_{10}$(M/H)–${\mathrm{log}}_{10}$(M/H)_⊙. In Figure 3, the lines of sight of De Cia et al. (2021) are represented as red arrows, whose sense indicates if their given R_G correspond to a lower or upper limit. The metallicities of the H ii regions are obtained from the O/H ratios given by Arellano-Córdova et al. (2021) subtracting $12+{\mathrm{log}}_{10}$(O/H)_⊙ = 8.69, the value of the solar O abundance used as a reference by De Cia et al. (2016, 2021), that corresponds to the photospheric O/H ratio recommended by Asplund et al. (2009). As we can see, the metallicity distributions of the two kinds of objects represented in Figure 3 are completely different. It is remarkable the extremely large spread of [M/H] values found by De Cia et al. (2021) between 8 and 9 kpc, a range of R_G that includes the maximum and minimum metallicities they report. De Cia et al. (2021) explicitly say that their observational data do not show evidence of a radial metallicity gradient, arguing that the narrow range of R_G covered by their line of sight may be the reason for that. However, the H ii regions represented in Figure 3 do show a gradient slope of −0.037 ± 0.019 dex kpc⁻¹ (represented by a dashed blue line in Figure 3), consistent with the one obtained by Arellano-Córdova et al. (2021) of −0.042 ± 0.009 dex kpc⁻¹ for their whole sample that covers R_G values between 4 and 17 kpc.

Zoom In Zoom Out Reset image size

In Table 1, we compare the metallicity variations found by De Cia et al. (2021) with those obtained by Arellano-Córdova et al. (2021) for H ii regions as well as other Population I objects. We give two different sets of values for De Cia et al. (2021) because they use two approaches for estimating neutral cloud metallicities, the "relative method" that provides dust-corrected abundances for individual metals, introduced by De Cia et al. (2016), and the "F*" method that gives the total metallicity, developed by Jenkins (2009). In the table, we give the mean and standard deviation of [M/H] (σ([M/H])) as well as these same magnitudes weighted by the inverse of the square of the uncertainty of each individual metallicity determination. In addition, we include in Table 1 the median of the uncertainty of [M/H] as well as the minimum and maximum [M/H] values of each data set and the difference between both quantities (${[{\rm{M}}/{\rm{H}}]}_{\max }\mbox{--}{[{\rm{M}}/{\rm{H}}]}_{\min }$). In the last two lines of the table, we include the mean difference between the individual [M/H] values with respect to the gradient fit—what we call [M/H] scatter—and the maximum amplitude of the differences of [M/H] measured above and below the gradient fit (Δ[M/H]) but only for those data sets where a gradient has been calculated. In each column of Table 1, we also indicate the element (O or Fe) that has been used as a proxy of metallicity in each reference. The corresponding [M/H] for each object has been determined by subtracting the solar O/H or Fe/H ratios adopted by De Cia et al. (2016, 2021). In the case of Fe, those authors use $12+{\mathrm{log}}_{10}$(Fe/H)_⊙ = 7.475, the mean of the photospheric and meteoritic solar abundances recommended by Asplund et al. (2009).

Table 1. Mean Metallicities and Variations in Different Young Population Objects

	Neutral Clouds						Classical	Young
	DC21		H ii Regions	B Stars			Cepheids	Clusters
	Relat. Met.	F* Met.	AC21	NP12		B19	L18	D20
Distance range	dHel < 3 kpc		dHel < 3 kpc	dHel < 0.5 kpc		dHel < 3 kpc	dHel < 3 kpc	RG = 5–11 kpc
Elements	Several		O	O		O	Fe	Fe
No. Objects	25		20	20		9	45	10
Mean [M/H]	−0.17	−0.24	−0.18	0.07	0.05	0.01	0.04	−0.04
σ([M/H])	0.29	0.22	0.09	0.05	0.02	0.06	0.09	0.10
Weighted Mean [M/H]	−0.49	−0.48	−0.18	0.06	0.06	−0.05	0.02	−0.14
Weighted σ([M/H])	0.27	0.23	0.06	0.05	0.02	0.07	0.07	0.11
Median [M/H] error	0.11	0.12	0.08	0.09	0.10	0.07	0.09	0.02
${\left[{\rm{M}}/{\rm{H}}\right]}_{\min }$	−0.78	−0.69	−0.35	−0.05	0.01	−0.13	−0.10	−0.28
${\left[{\rm{M}}/{\rm{H}}\right]}_{\max }$	0.28	0.11	−0.04	0.15	0.08	0.08	0.28	0.12
${\left[{\rm{M}}/{\rm{H}}\right]}_{\max }-{\left[{\rm{M}}/{\rm{H}}\right]}_{\min }$	1.06	0.80	0.31	0.20	0.07	0.21	0.38	0.40
Quantities defined with respect to the gradient fit determined in each reference
$\left[{\rm{M}}/{\rm{H}}\right]$ scatter	⋯	⋯	0.06	⋯	⋯	0.03	0.06	0.06
${\rm{\Delta }}\left[{\rm{M}}/{\rm{H}}\right]$	⋯	⋯	0.27	⋯	⋯	0.16	0.43	0.22

Note. DC21: De Cia et al. (2021); AC21: Arellano-Córdova et al. (2021); NP12: Nieva & Przybilla (2012); B19: Bragança et al. (2019); L18: Luck (2018); D20: Donor et al. (2020).

Download table as:  ASCIITypeset image

In Figure 4, we show some quantities included in Table 1 for a more graphic comparison. We represent the mean of the [M/H] values (horizontal lines)—not the weighted mean—as well as their σ([M/H]) (colored rectangles) and the maximum and minimum values of [M/H] of the different data sets. The length of the solid vertical lines delimited by triangles corresponds to the [M/H]_max–[M/H]_min quantity given in Table 1.

Zoom In Zoom Out Reset image size

In Table 1 and Figure 4, we can note that the mean metallicity of neutral clouds and H ii regions is fairly similar. However, it seems remarkable that since the mean and weighted mean of [M/H] are almost the same for H ii regions, they differ by a factor of 0.24 or 0.32 dex (depending on the method for correcting for dust depletion used) in the case of neutral clouds. This indicates that the locus of neutral clouds with the best metallicity determinations are displaced to low [M/H] values. From Table 1 and Figures 3 and 4 it is evident that σ([M/H]) and [M/H]_max–[M/H]_min values are dramatically different for neutral clouds and H ii regions. It is also remarkable that while σ([M/H]) is of the order of the median uncertainty of the individual determinations for H ii regions, both quantities differ in the case of neutral clouds. This indicates that the metallicity distribution of neutral clouds is further from a normal distribution than that of H ii regions. As expected, for H ii regions, the scatter of [M/H] with respect to the linear fit to the gradient is smaller than σ([M/H]).

3.1. Dust and O Abundance Determination in H ii Regions

In addition to neutral clouds and H ii regions, Table 1 and Figure 4 also include metallicity variations derived for different kinds of Population I objects belonging to the local Galactic thin disk: early B-type stars, classical Cepheids, and young open clusters. Due to their short lifetimes, all these stellar objects are still located not far from their birthplace and are unaffected by enrichment processes due to stellar nucleosynthesis or depletion onto dust grains. Therefore, their metallicity distribution should be representative of the present-day local ISM. The metallicity values of the different objects included in Table 1 or Figure 4 are calculated with respect to the solar O or Fe abundances adopted by De Cia et al. (2016, 2021).

The metallicities of Nieva & Przybilla (2012) included in Table 1 and Figure 4 are based on photospheric abundances of O and Fe of 20 bright, single, sharped lined, and chemically inconspicuous early B-type stars located in the solar neighborhood, at heliocentric distances <500 pc. Their observational data consist of high-resolution and high-S/N spectra taken with different spectrographs that are analyzed by applying non-LTE line-formation computations in a self-consistent way. In Table 1 and Figure 4, we also present data obtained by Bragança et al. (2019), the most recent study of Galactic radial abundance gradients using young main-sequence OB stars. They use a sample of 31 stars located in the outer Galactic disk and perform a non-NLTE analysis of high-resolution spectra obtained with the MIKE spectrograph on the Magellan Clay 6.5-m telescope at Las Campanas Observatory. Bragança et al. (2019) find a linear metallicity gradient with a somewhat steeper slope than that obtained by Arellano-Córdova et al. (2020, 2021) for H ii regions but consistent with the recalculation made by Méndez-Delgado et al. (2022) assuming t² > 0 (see Section 4.1). Since the sample of Bragança et al. (2019) covers stars located at R_G between 8.4 and 15.6 kpc, we only consider those overlapping the area covered by the observations of De Cia et al. (2021), corresponding to stars with heliocentric distances <3 kpc. As we can see in Table 1 and Figure 4, the mean [M/H] values obtained from the data used for B-type stars are clearly larger than those obtained for neutral clouds or H ii regions, but their σ([M/H]) and [M/H]_max–[M/H]_min values, are far more consistent with those obtained from H ii regions than from neutral clouds. There are other metallicity determinations in early-type stars available in the literature. From determinations of the O/H ratio for a sample of 13 B-type stars of the Ori OB1 association, Simón-Díaz (2010) finds a σ([M/H]) and[M/H]_max–[M/H]_min of 0.04 dex and 0.10 dex, respectively. In fact, this author highlights the high degree of chemical homogeneity of the Ori OB1 stars. Finally, Lyubimkov et al. (2013) derive stellar abundances of 22 B-type stars with heliocentric distances ≤600 pc. They find metallicity variations somewhat larger than the other cited references: σ([M/H])=0.13 dex and [M/H]_max–[M/H]_min = 0.39 dex, but still far from the larger variations shown by neutral clouds.

Classical or Population I Cepheids are variable stars with masses between 4 and 20 M_⊙ (Turner 1996), which correspond to ages ≲300 Myr (Bono et al. 2005). From an LTE analysis of high-resolution spectra, Luck (2018) determined Fe abundances for several hundreds of Galactic classical Cepheids located at R_G values from 3 to 18 kpc, the largest compilation of abundances of classical Cepheids to date. Luck (2018) finds a linear gradient fit with a slope very similar to that obtained for H ii regions. Of the sample studied by that author, we only consider the results for a subsample of 52 Cepheids with the highest quality abundance determinations based on observations covering five or more phases of the variability period, of those we select the 45 objects located in the Galactic area covered by the observations of De Cia et al. (2021), having heliocentric distances <3 kpc. Table 1 and Figure 4 show that the metallicity variations—parameterized with σ([M/H]) and [M/H]_max–[M/H]_min—found for local classical Cepheids are similar to those obtained for H ii regions. Korotin et al. (2014) determined the radial O abundance gradient from data of several hundreds of Cepheids, most of them Type II Cepheids, old low-mass Population II giants, with ages of about several Gyr. Those authors find σ([M/H]) and [M/H]_max–[M/H]_min values of 0.15 and 0.42 dex, respectively, somewhat larger than the variations obtained for classical Cepheids and H ii regions—as expected because their much larger ages, but still clearly smaller than the numbers obtained by De Cia et al. (2021) for neutral clouds.

In the last column of Table 1 and in Figure 4, we include the metallicity parameters for open clusters of the OCCAM (Open Cluster Chemical Abundances and Mapping) survey published by Donor et al. (2020). This work makes use of infrared-based spectroscopic data from the APOGEE-2/SDSS-IV DR16 (Jönsson et al. 2020) of more than a hundred open clusters. In our case we have selected a subsample containing the 10 youngest clusters—with ages <400 Myr—flagged as "high quality" in the database and located in the interval of R_G between 5 and 11 kpc, consistent with the area covered by the observations by De Cia et al. (2021). The values of σ([M/H]) and [M/H]_max–[M/H]_min for these clusters, 0.10 and 0.40 dex, respectively, are consistent with those of the other Population I objects and H ii regions included in Table 1.

A final aspect we want to highlight from the data gathered in Table 1 is the remarkable consistency of the values of the scatter of [M/H] around the gradient fit and Δ[M/H]_max obtained for the different kinds of Population I objects. This strongly suggests that the true degree of homogeneity of the local ISM should be not far from those numbers.

In Section 2.2, we commented on the results by Domínguez-Guzmán et al. (2022), showing the remarkable homogeneity of the metallicity in the Magellanic Clouds. The almost constant O abundance they find in the LMC contrast with the results obtained for neutral clouds by Roman-Duval et al. (2021). These last authors determine gas-phase metallicities from Hubble Space Telescope UV spectra in the line of sight of 32 massive stars of the LMC, finding that the metallicity increases 0.8 dex from the west to the east side of the galaxy. Considering that the H ii regions observed by Domínguez-Guzmán et al. (2022) also cover both sides of the LMC, the apparent metallicity distribution found by Roman-Duval et al. (2021) should be related not to true gas-phase variations but to differences in the fraction of dust depletion along the LMC disk. In fact, this last possibility is also considered by Roman-Duval et al. (2021), who interpret the variation of dust depletion as a likely product of the increased feedback from star formation and the turbulence induced by the collision with the Milky Way.

4.1. The Abundance Discrepancy and Metallicity in H ii Regions

The comparison of the data compiled in Table 1 and Figure 4 provides a fairly clear picture: the amplitude of metallicity variations found in H ii regions, B-type stars, classical Cepheids, and young clusters are similar and relatively small, implying that the ISM is basically chemically homogeneous in the azimuthal direction of the Galactic disk around the Solar neighborhood, consistently with an efficient gas mixing, which is also the prediction by theoretical works (i.e., Edmunds 1975; Roy & Kunth 1995; de Avillez & Mac Low 2002; Yang & Krumholz 2012), where metallicity inhomogeneities are wiped out in the azimuthal direction in less than a Galactic orbital period (i.e., Petit et al. 2015). As Roy & Kunth (1995) discussed, there are several mixing mechanisms in the ISM showing different timescales that increase as the spatial scale increases. At large spatial scales, from 1–10 kpc, cloud–cloud collisions in a shear flow due to differential rotation produce effective mixing in the azimuthal direction at timescales ≤10⁹ yr. For intermediate spatial scales, 0.1–1 kpc, those cloud–cloud collisions in addition to expanding interstellar bubbles—powered by supernova explosions and stellar winds—are combined with the shear flow to produce mixing with shorter timescales, around 10⁸ yr, of the order of the half-revolution time of the spiral pattern at the Sun. At smaller spatial scales, 1–1000 pc, turbulent diffusion in ionized, neutral, and molecular clouds have different timescales from 10⁷ to 10⁹ yr, but fluid instabilities in H ii regions may be as short as 10⁶ yr.

Considering that disk rotation is the main agent for mixing in spiral galaxies and the effect of selective loss of metals due to galactic winds, Roy & Kunth (1995) suggested that dwarf galaxies—with weaker rotation fields and lower masses—should show larger metallicity variations. However, as we discuss in Sections 2.2 and 3, this is not confirmed by the observations in dwarf irregular galaxies and in the most recent abundance determinations of H ii regions in the Magellanic Clouds by Domínguez-Guzmán et al. (2022), who find a highly homogeneous metallicity distribution along their disks, indicating the apparent large efficiency of mixing processes even in dwarf irregular galaxies.

Comparing results for neutral clouds inside the Orion and the Ophiucus areas, De Cia et al. (2021) make estimates of the minimum physical scale of the metallicity variations they observe, finding that they are of the order of tens of parsecs or even down to a few parsecs. In Figure 2 we can see a group of eight H ii regions rather close to each other located at Galactic longitudes between 6° and 19° and covering heliocentric distances between 1.15 and 1.91 kpc. The mean distance between the H ii regions of this group is ∼150 ± 60 pc and their σ([M/H]) and [M/H]_max–[M/H]_min values are determined from the data obtained by Arellano-Córdova et al. (2020, 2021) are 0.07 and 0.22 dex. These numbers are very similar to those obtained for the whole subsample of H ii regions of Arellano-Córdova et al. (2020, 2021) included in Table 1. This result is consistent with the interpretation that much of the metallicity variations that we find in H ii regions may be related to the quality of the observations and uncertainties in abundance determinations.

It is well known that current chemical evolution models explain the presence of radial abundance gradients in spiral galaxies in the context of the inside-out formation paradigm. The galactic disks grow by means of radially dependent gas infall from the halo that drives star formation (i.e., Tosi 1988; Matteucci & Francois 1989; Boissier & Prantzos 1999; Chiappini et al. 2001). This is a strong argument that the Milky Way has accreted—and is still accreting—low-metallicity gas to maintain its current star formation activity. De Cia et al. (2021) interpret the large metallicity variations they found in the neutral clouds of the thin disk as the product of the falling of high-velocity clouds ⁴ (HVC) of pristine gas with zero depletion onto the Milky Way. Necessarily, in this scenario, those clouds should not mix efficiently with the ISM until they reach the thin disk in order to maintain their low metallicity. However, the theoretical and observational evidence available does not seem to support that scenario.

Three-dimensional hydrodynamical simulations of cloud–interstellar medium interactions predict that mixing between the cloud and its surrounding gas is very efficient, raising the metallicity of the HVC as it crosses the Galactic halo (Gritton et al. 2014; Henley et al. 2017; Heitsch et al. 2022). Presumably, since the interaction begins long before the clouds slow to the velocity of the Galactic ISM and dissipate, this increase in metallicity should be very important when the HVC reaches the thin disk. In fact, as the most recent calculations made by Heitsch et al. (2022) predict, contamination of an HVC traveling through the halo may be so large that the final hydrogen mass may contain a fraction less than 10 per cent of the original cloud material. As Heitsch et al. (2022) remark, in all likelihood, metallicity measurements in HVCs will not represent original cloud properties, and surely they will be much higher when those HVCs reach the thin disk. Observed [M/H] values in HVCs, like the Magellanic Stream or Complex C, located at distances of ∼50 and 10 kpc, respectively, range between −1.0 and −0.3 dex (Wakker et al. 1999; Fox et al. 2010; Shull et al. 2011; Richter et al. 2013; Richter 2017). We can see that those metallicities are similar to the ones reported by De Cia et al. (2021), but the distances to the Galactic disk of HVCs are far larger than the distances to the local neutral clouds observed by De Cia et al. (2021), which are located at heights ≲250 pc above the Galactic plane, ∼2/3 of them lying within 100 pc. Moreover, considering that ∼90 percent of the accretion rate of the Milky Way has its origin in the Magellanic System (Richter 2017), one would expect that the population of neutral clouds coming from such accretion at the Galactic plane should show metallicities larger than those of the Magellanic Clouds—[M/H] = −0.32 and −0.68 for LMC and SMC, respectively (Domínguez-Guzmán et al. 2022)—due to mixing as they fall and decelerate.

Intermediate-velocity clouds (IVC) have ∣v_LSR∣ ≈ 30–100 km s⁻¹ and tend to be at heights between 0.5 and 1 kpc above the Galactic plane (Wakker 2001). They are believed to originate in the Galactic Fountain or from HVCs coming from the intergalactic medium slowed down as they fall toward the Galactic plane. IVCs show metallicities between −0.3 and 0.0 dex (i.e., Richter et al. 2001a, 2001b; Wakker 2001; Richter 2017), higher than those of HVCs. Cloud IV21, which lies 300 pc above the disk, is the IVC with the lowest known metallicity. Hernandez et al. (2013) measure [M/H] = −0.43 ± 0.12 dex for this object, proposing that IV21 is a decelerated infalling low-metallicity HVC that is mixing with disk gas in the lower Galactic halo. About 1/3 of the sightlines observed by De Cia et al. (2021) show metallicities lower than that value despite their lower heights with respect to the Galactic plane.

Considering all the aforementioned theoretical and observational considerations about the origin, distance distribution, initial metallicities, and mixing processes of infalling HVCs and IVCs, we consider it unlikely that they may be the origin of a significant population of low-metallicity neutral clouds in the local Galactic thin disk.

In this paper we discuss and confront recent results about metallicity variations in the local Galactic ISM obtained from H ii regions and neutral cloud observations. While the data by Arellano-Córdova et al. (2020, 2021) indicate that the gas in H ii regions is fairly well mixed at a given Galactocentric distance, De Cia et al. (2021) report the finding of gas-phase metallicity variations up to a factor of ten in neutral clouds along the line of sight of massive stars located within 3 kpc around the Sun. We compare these discrepant results with high-quality recent metallicity determinations of other objects that are considered tracers of the chemical composition of the present-day ISM: B-type stars, classical Cepheids, and young clusters. We find that the metallicity variations obtained for these last kinds of objects are consistent with each other, similar to that obtained for H ii regions, but significantly smaller than those obtained for neutral clouds. Moreover, recent direct abundance determinations in H ii regions of the Magellanic Clouds and other nearby spiral galaxies obtain metallicity variations of the order of the errors of individual abundance determinations, in agreement with the Milky Way results obtained by Arellano-Córdova et al. (2020, 2021).

The hypothesis of a large population of low-metallicity clouds in the Galactic thin disk around the Sun proposed by De Cia et al. (2021) is also in contradiction with the prediction of theoretical studies of gas mixing. These studies agree that metallicity inhomogeneities are wiped out in the azimuthal direction of the disk in less than a Galactic orbital period. On the other hand, that scenario does not fit the picture provided by observational results about metallicity distribution in HVCs falling onto the Galactic plane, as well as models about the mixing of HVCs with the surrounding ISM. We expect that the final products of decelerated HVCs when reaching the Galactic plane should produce a population of clouds with larger metallicities than those found by De Cia et al. (2021).

The disagreement between the amplitude of the metallicity variations found in neutral clouds and the rest of the objects have two possible explanations: (a) one of the two results is incorrect or (b) if both are correct, the gas mixing in the thin disk is mainly produced just at the onset of star formation, in the transition phase between neutral cloud to H ii region. If we consider the first explanation, it is very difficult to explain why four kinds of objects: H ii regions, B-type stars, classical Cepheids, and open clusters give similar results—even similar gradients—considering the completely different techniques used to derive metallicities in them. Only the results on neutral clouds by De Cia et al. (2021) break an otherwise coherent picture, so it is reasonable to think that the problem lies in this latter data set, perhaps related to an incorrect correction for dust depletion. The second possibility, that the gas mixing is produced at the onset of star formation seems rather unlikely. This hypothetical process would require a quite short timescale that will imply a short spatial scale for mixing. Moreover, a process with such limitations would probably be unable to produce the small metallicity variations we see in all the objects—apart from neutral clouds—at scales of several kiloparsecs. In any case, if the two results are correct, neutral clouds have very large metallicity variations and H ii regions and young stars do not, this would imply that the presence of a significant population of low-metallicity clouds definitely does not have any practical influence on the subsequent composition of the stellar population and, therefore, on the chemical evolution of the Galactic disk.

We acknowledge support from the Agencia Estatal de Investigación del Ministerio de Ciencia e Innovación (AEI-MCINN) under grant Espectroscopía de campo integral de regiones H ii locales. Modelos para el estudio de regiones H ii extragalácticas with reference 10.13039/501100011033. JEM-D thanks the support of the Instituto de Astrofísica de Canarias under the Astrophysicist Resident Program and acknowledges support from the Mexican CONACyT (grant CVU 602402). JG-R acknowledges support from an Advanced Fellowship under the Severo Ochoa excellence program CEX2019-000920-S. The authors acknowledge support under grant P/308614 financed by funds transferred from the Spanish Ministry of Science, Innovation and Universities, charged to the General State Budgets and with funds transferred from the General Budgets of the Autonomous Community of the Canary Islands by the MCIU.