A New Search for Star Forming Regions in the Southern Outer Galaxy

Star-formation in the outer Galaxy is thought to be different from the inner Galaxy, as it is subject to different environmental parameters such as metallicity, interstellar radiation field, or mass surface density that all change with Galactocentric radius. We therefore aimed at getting a more detailed view on the structure of the outer Galaxy, determining physical properties for a large number of star forming clumps and understanding star-formation outside the Solar circle. We use pointed $^{12}$CO(2-1) observations conducted with the APEX telescope to determine the velocity components towards 830 dust clumps identified from 250 $\mu$m Herschel/Hi-GAL SPIRE emission maps in the outer Galaxy between $225\deg<\ell<260\deg$. We determined kinematic distances from the velocity components, in order to analyze the structure of the outer Galaxy and to estimate physical properties such as dust temperatures, bolometric luminosities, clump masses, and H2 column densities for 611 clumps. We find the CO clouds to be strongly correlated with the highest column density parts of the Hi emission distribution, spanning a web of bridges, spurs and blobs of star forming regions between the larger complexes, unveiling the complex three-dimensional structure of the outer Galaxy in unprecedented detail. Using the physical properties of the clumps, we find an upper limit of 6% (40 sources) to be able to form high-mass stars. This is supported by the fact that only 2 methanol Class II masers or 34 known or candidate Hii regions are found in the whole survey area, indicating an even lower fraction to be able to form high-mass stars in the outer Galaxy. We fail to find any correlation of the physical parameters of the identified (potential) star forming regions with the expanding supershell, indicating that although the shell organizes the interstellar material into clumps, their properties are unaffected.


Introduction
Star formation and the processes involved with it, such as disc accretion, molecular outflows, stellar winds, chemical enrichment and energy input into the local environment from radiation, mechanical energy from the outflows, and supernova explosions of the most massive stars, play an important role in determining the structure of the interstellar medium (ISM) and driving the evolution of a galaxy (Kennicutt 2005). As star formation takes place in molecular clouds, it is not only important to know the distribution of these clouds within the Galaxy but also how different environmental conditions affect their properties, structure and dynamics. In the Milky Way, the distribution of molecular clouds as traced by CO (García et al. 2014, Rice et al. 2016, Miville-Deschênes et al. 2017) shows a strong peak within ∼2 kpc of the Galactic center, and another peak at a distance of ∼5 kpc from the Galactic center, after which the distribution drops off out to ∼20 kpc. The vast majority of the molecular gas found in the Galaxy is located in the inner Galaxy (i.e. ∼85% within the Solar circle at R < 8.3 kpc; Miville-Deschênes et al. 2017) and as a result most of the previous studies have focused on star formation in the inner part of the Milky Way. Many thousands of low-and high-mass star forming regions have been identified and investigated by a large community (e.g. Mooney & Solomon 1988, Brand & Blitz 1993, Urquhart et al. 2014a, Elia et al. 2017. In contrast to the molecular clouds in the inner Galaxy, the clouds located in the outer Galaxy (i.e. outside the Solar circle at R > 8.3 kpc) contribute only ∼15% of the total molecular gas of the Milky Way (Miville-Deschênes et al. 2017). However, observations towards the outer Galaxy do not suffer from the kinematic distance ambiguities that plague studies of the inner Galaxy (Roman-Duval et al. 2009, Wienen et al. 2015 and source confusion is significantly reduced due to a lower density of molecular clouds. There are, therefore, a number of observational advantages to studying molecular clouds located outside the Solar circle. The physical conditions are also very different compared to those found in the inner Galaxy (e.g., the Hi density decreases, UV radiation field is less intense, and the general cosmic-ray flux and metallicity decreases with increasing Galactocentric distance (e.g. Bloemen et al. 1984, Rudolph et al. 1997 Article number, page 1 of 21 arXiv:2010.11049v1 [astro-ph.GA] 21 Oct 2020 A&A proofs: manuscript no. paper and so studies of these objects provide valuable insight into how the initial conditions of the clouds, and the physical conditions of their local environment, affect star formation. There have been a number of studies that have investigated the distribution and properties of molecular clouds in the outer Galaxy (e.g. Wouterloot & Brand 1989, May et al. 1997, Heyer et al. 2001, Nakagawa et al. 2005, Elia et al. 2013. These studies found the clouds to be, in general, smaller, less massive and have smaller line-widths compared to clouds located in the inner Galaxy (e.g. Dame et al. 1986, Solomon et al. 1987). In addition, the conditions of molecular clouds located in different spiral arms might be different (e.g. Benjamin et al. 2005). Especially the impact of the entry shock experienced from the material entering a spiral arm should be stronger within the solar circle, as the surface density of the ISM drops by about an order of magnitude around the solar circle (Heyer & Dame 2015), and hence the amplitude of the spiral density wave is significantly smaller in the outer Galaxy. Furthermore, as the velocity difference between the ISM and the spiral pattern reverses sign around (and reaches zero at) the co-rotation radius R cr ≈ 10.9 kpc (Koda et al. 2016) in the outer Galaxy, the entry shock should be further diminished. Therefore, outside the solar circle radius supernovae explosions may take over as the dominant mechanism determining the state of the ISM (e.g. Kobayashi et al. 2008). In this paper we will build on these previous studies and investigate the properties of a large sample (∼800) of molecular structures located in the Galactic longitude range 225 • < < 260 • . This region includes a large section of the Perseus and Outer arm and so will allow us to compare the properties of molecular clouds located in the different arms and inter-arm regions (e.g. Eden et al. 2013;2015). This study has also been designed to complement the recent studies of the inner Galaxy reported by Urquhart et al. (2018). In an upcoming paper (from here on Paper II) we will also compare the results of the two studies, allowing us to investigate trends in the clump properties and their distribution over the whole range of Galactocentric distances out to ∼ 16 kpc. In this paper, our main goals are to analyze the structure of the Galaxy outside the Solar circle, to identify possible complexes of star formation and characterize and compare their physical properties. We use pointed 12 CO(2-1) observations towards molecular clouds and the rotation curve from Brand & Blitz (1993) to calculate kinematic distances from the observed radial velocity ( lsr ). Archival dust continuum emission maps are used to determine physical properties, following the methods developed in our previous work for the inner Galaxy (König et al. 2017, Urquhart et al. 2018. The paper is structured as follows: in Sect. 2 we describe how we selected our sources, the setup for the 12 CO(2-1) observations, how we obtained velocities and determined Galactocentric distances. In Sect. 3 we present the physical properties obtained from dust spectral energy distributions, using the distances we determined. We discuss these properties and the star forming relations for a detailed look on star formation in the outer Galaxy. In Sect. 5 we will discuss the large-scale structures encountered in this part of the outer Galaxy and investigate the influence of the supershell on its environment. In the final section we give an overview of our findings as well as an outlook on our future work and upcoming papers.

Molecular line data used
For the present work we identified and selected dust continuum sources from the Herschel Hi-GAL continuum maps and determined distances for selected sources using dedicated observations of the associated 12 CO(2-1) emission. Table 1. Astrometric data and integrated flux as determined by SExtractor for the extracted clumps for the first 20 clumps from a total of 23,817 sources. Columns are as follows: name of the source, Galactic longitude, Galactic latitude, full-width-at-half-maximum source size and the 250 µm integrated flux as measured by SExtractor.

Source extraction and selection
The region in the outer Galaxy chosen for this survey (225°≤ ≤ 260°, −3°≤ b ≤ 0.5°) was selected, as this part of the outer Galaxy has never been observed with high spatial resolution or sensitivity before. Furthermore, the sections of the spiral arms in this region are well separated in velocity, making it relatively straightforward to associate objects with their parent spiral arm. The latitude range was selected as the minimum and maximum latitudes covered by the Herschel Hi-GAL dust continuum emission maps to ensure availability of complementary data. The Herschel SPIRE 250 µm emission data obtained by Hi-GAL (Molinari et al. 2010) was used to identify clumps in this region. Note that due to the Galactic warp the latitude range covered by Hi-GAL is centred around b = −1°spanning ∼2.5°. Although the authors are aware of the Hi-GAL compact source catalogues that are publicly available for the inner Galaxy (Molinari et al. 2016), their outer Galaxy counterpart is not released yet. Therefore an independent method was chosen to obtain a source catalogue for the targeted area of the outer Galaxy. To identify emission peaks and obtain their source sizes, SExtractor (Bertin & Arnouts 1996) was used, as it was used by team members in a similar way to produce the ATLASGAL compact source catalogue (Contreras et al. 2013, Urquhart et al. 2014c with great success, and will allow for more reliable comparisons between the outer Galaxy and ATLASGAL datasets. As a general approach we searched for emission peaks that have at least 4 pixels (i.e. ∼1 beam) above a threshold of 3σ rms above the local background noise level. We used two different background mesh sizes of 64 or 32 pixels to either exclude or include the 6 brightest and most complex regions of emission, respectively, as these crowded regions would introduce a bias in the source selection process. As a result, we obtained two catalogues: one excluding the brightest regions, which includes  12,783 sources, and a second one that was optimized to identify clumps located in the bright regions, that contains 15,874 sources. Merging both catalogues and removing duplicate entries, we obtained a full catalogue identifying a total of 23,817 sources in the 225°≤ ≤ 260°region. The result of the source extraction from the SPIRE 250 µm images can be seen in Fig. 1 which includes a variety of bright complexes and dark regions devoid of emission. In Table 1 we give the parameters of the extracted sources for a small portion of the catalogue. As one major goal of this research is to probe the full distance range from the nearest to the farthest sources, we used the subset excluding the 6 brightest regions to select the majority of our sources. However, in order to characterize these regions, too, we manually added 34 emission peaks from the full catalogue, ensuring these regions are included but not over-represented (see Table 2 and Fig. 2 for an overview of these 6 regions). For the observations outside the six brightest and most complex regions, we selected the 100 sources yielding the highest peak flux as well as the 100 sources with the highest integrated flux, accounting for 196 sources with 4 sources satisfying both criterias. An additional 587 sources were finally selected randomly from the brightness limited subset, picking up faint sources that are more likely to be located at farther heliocentric distances. In total, 830 sources of the extracted emission peaks were observed. To show that our source selection procedure does not introduce a strong flux bias, we show in Fig. 3 the histogram of the aperture flux at 250 µm of all extracted sources (including the brightest regions) together with the histogram of the aperture flux of the observed sources. As can be seen, our observed sources sample the whole range of the flux distribution of the complete catalogue very well down to the limit of ∼2.5 Jy, which was imposed as a selection threshold to ensure the 12 CO(2-1) line could be detected at a 3σ level in a reasonable amount of integration time ( 3 min).

Observations
Carbon monoxide (CO) is the second most abundant molecule in the Milky Way after H 2 . It is widely found in molecular clouds, and even more so in the denser parts constituting the dusty clumps we are investigating in this work. Observing the CO emission towards a dust clump allows us to obtain a velocity for a given clump and from this infer a distance to the clump. We used the facility receiver APEX-1 (SHeFI; Vassilev et al. 2008) at the Atacama Pathfinder Experiment 12 m submillimeter telescope (APEX; Güsten et al. 2006) to obtain line-of-sight velocities ( lsr ) of the selected sources. Two wide-band Fast Fourier Transform Spectrometers (FFTS; Klein et al. 2012) make up the back-ends, each consisting of 32,768 spectral channels covering an instantaneous bandwidth of 2.5 GHz. We targeted the 12 CO(2-1) transition at 230.5 GHz, as line-of-sight confusion is not as much an issue for the outer Galaxy than for the inner Galaxy, and the higher column densities allow for detection of sources at greater distances with similar integration times when compared to the rarer and less bright 13 CO(2-1) and C 18 O(2-1) transitions. With this setup we were able to observe the targeted transition at a velocity resolution of ∼ 0.1 km s −1 . The observa-Article number, page 3 of 21 A&A proofs: manuscript no. paper  tions towards 830 sources were conducted between July 2013 and December 2016 as a bad weather backup project at APEX with precipitable water vapour (PWV) up to 6 mm. To reach an rms < 0.1 K per channel, each source was observed in position switching mode for 3-6 min, depending on PWV, elevation and line-strength. The off-positions were selected as relative offsets with a separation of 1°roughly perpendicular to the Galactic plane. The pointed 12 CO(2-1) observations were obtained with an average PWV of 2.7 mm, resulting in an average RMS of 0.05 K at a smoothed channel resolution of 1 km s −1 . Sample spectra are shown in Fig. 4. In total, we found 37 spectra without any emission, leaving 759 (∼91%) spectra for further analysis. A summary of all observational parameters is given in Table 3. To obtain CO(2-1) velocity information for each clump, we applied the methods described in the following section.

Identifying velocity components
The obtained spectra were reduced using the Continuum and Line Analysis Single-dish Software (CLASS 1 ). To obtain the velocity components from the observed spectra we first combined all scans for a single observed position into a single spectrum. This was successively smoothed to a velocity resolution of 1 km s −1 and a linear baseline was subtracted. The spectra were then subjected to a Python code, where they were limited to the velocity range of −20 km s −1 < lsr < 150 km s −1 , in order to limit the spectra to velocity range expected for the outer Galaxy. Coherent groups of emission and absorption likely associated with a single cloud were determined by defining a window where all emission is above the 3σ noise level. These emission groups are then fitted iteratively, starting with the brightest group. First the spectrum is de-spiked, after which the number of peaks in a group is determined and Gaussian profiles are fitted to the peaks, and the resulting fit is subtracted from the original spectrum. The procedure is then repeated until no residual emission above 3σ is found and all groups are fitted. In order to avoid adding too many emission components that are close to each other and are likely associated with the same cloud, we only considered peaks separated by at least twice the width of the fitted Gaussian as major emission components. The same process was repeated for negative emission features, allowing us to identify observations with a contaminated off-position. In Fig. 4 we show example spectra of velocity measurements obtained towards sources located in the outer Galaxy. In the upper left panel a single emission component along the line-of-sight can be fitted with a single Gaussian profile. This allows the velocity to be immediately assigned to the dust clump without further analysis. The situation becomes more difficult when a cloud is either composed of multiple components or there are multiple clouds located along the line of sight (Fig. 4, upper right panel). Furthermore, as CO is the second most abundant molecule in the Milky Way, the reference position used when taking the spectrum might be contaminated, resulting in negative features in the spectrum (Fig. 4, lower left panel). Contaminated spectra pose the problem that the emission might be at a similar velocity as the negative feature, thus rendering the velocities less reliable. In extreme cases all these three effects are present in a single spectrum taken (Fig. 4, lower right panel). Although these effects complicate the analysis, in most cases still a velocity can be assigned to the corresponding clump. Assuming that the emission as seen in the continuum maps is associated mainly with the brightest emission found in CO, only the CO emission with the highest integrated intensity is taken into account. If this CO cloud has an integrated intensity at least twice as high as the rest of the CO emission, we assume that the dust emission is associated with it. In case a negative feature is located close to it (dv < 1 km s −1 ), the uncertainty of the velocity measurement is increased by the width of the negative feature from the off-position, considering the velocity measurement is still usable but with a higher uncertainty. With the line-of-sight velocity of a clump known, we can determine its distance from the rotation curve of the Galaxy. As there is only one rotation direction in the southern outer Galaxy, namely from higher to lower longitudes, no distance ambiguity exists (in contrast to the inner Galaxy where this poses a problem), allowing for a direct conversion from the observed line-of-sight velocities to heliocentric distance. We were able to fit spectra for 759 (91%) of the observed 830 lines-of sight, identifying a total of 982 clouds consisting of 1102 velocity components from the CO(2-1) emission above the 5σ level. From here on we refer to a single fitted Gaussian as a velocity component and to any coherent group of emission as a cloud, as we assume that these groups of emission are physically connected with similar velocities but distinct components. We found the source-off reference positions, selected as a relative offset from the target position of one degree, to be contaminated for 331 sources of the sample (39.9%), as shown for two lines-of-sight in the lower panels of Fig. 4. As the uncertainty in distances determined from the lsr are mostly dominated by the uncertainty in the rotation curve used for the calculation rather than from the velocity, we assume that an uncertainty in lsr of a few km s −1 due to contamination in the source-off position, does not pose a serious problem. In fact, we conclude that the CO emission at the source-off position can be used to increase the number of positions for velocity measurements towards the outer Galaxy, but care has to be taken on their usage as discussed in the following paragraph. Therefore we also analyse the emission at the off positions when a source shows contamination. To do so, we simply invert the baseline corrected spectra and apply the same procedure as for the emission of uncontaminated sources, yielding velocity components for the offposition. We want to stress that these additional CO(2-1) clouds do not necessarily coincide with the peak of any dust clump or cloud, but as these components are clearly above the background noise level, they improve our view on the large-scale structures in the outer Galaxy. In contrast, for the SED analysis of the dust clumps in Section 3.1, we only take into account off-source positions that are matched with a SExtractor source position within one beam size (i.e. 30 ), yielding an additional 40 sources for which we can derive physical properties. If an absorption feature was found within 10 km s −1 of an emission peak, the uncertainties for all derived properties were increased by a factor of two (i.e. velocity, peak temperature and line-width). We choose a rather large range of 10 km s −1 , in order to make sure that we also cover cases with broad emission complexes, where the emission between two peaks cancels each other out. Analysing the emission at the off positions for the 331 contaminated spectra, we were able to add another 295 clouds above the 5σ noise level from 242 lines-of-sight to our analysis. This increases the total number of clouds by 23.1% from 982 to 1277 consisting of a total of 1415 components and covering a total of 1090 linesof-sight (targeted + reference). We present the results for each position in Table 4. As observations were often conducted under bad-weather conditions with high and fast varying PWV sometimes in excess of 5 mm, we calculate the main-beam brightness temperature only for sources that have a calibration within a certain time-frame according to the weather conditions 2 , resulting in the removal of 23 lines-of sight (29 clouds). For the remaining sources the main beam temperature T mb was then calculated from the antenna temperature T * A multiplying with a forward efficiency of η f = 95% and dividing by a beam efficiency of η mb = 75%. In case no calibration was found within the given time-frame, we do not give a main beam temperature or intensity. For 966 (88.6%) lines-of-sight only a single cloud is found, whereas multiple clouds are identified for 87 (11.4%) lines-ofsight. For 37 lines-of-sight no emission was found. In summary we identified a total of 1248 clouds for all 1090 lines-of-sight, yielding an average of 1.2 clouds per line-of-sight. We give all clouds (i.e. groups of emission) in Table 5, which we define as continuous emission features above 3σ rms, consisting of one or more fitted Gaussians. An example of two components (possibly arising from self absorption) making up one cloud can be seen in Fig. 4 (lower right) at ∼80 km s −1 indicated by the solid black bar at the lower part of the figure. The velocity found for the brightest velocity complex along a given line-of-sight is later used to assign a distance to the clump as seen in continuum emission, from which the physical parameters are derived.

Source velocities
In Fig. 5 we present the longitude-velocity ( -v) plot covering the observed field. The velocity components above the 5σ rms noise level for all positions are overlaid on the 12 CO(1-0) emission data cube from Dame et al. (2001) integrated for −3 deg ≤ b ≤ 0.5 deg as approximately covered by our sample, that is following the Galactic warp. The loci of the spiral arms as well as the local emission are indicated by the solid line (see Sect. 5.1 for details). As can be seen, although we randomly sampled, components obtained for the CO(2-1) line trace all the major structures found in the CO(1-0) emission map very well. In addition, due to the 15 times higher resolution of our pointed observations when compared to the data from Dame et al. (2001; i.e. 7.5 vs. 30 ), we are able to trace small structures that have been previously missed, as these would fall below the detection limit due to beam dilution or blending with other clumps within the beam. For example, the structures with the highest velocity components at any given longitude would trace the most distant arm. The structure traced by our pointed observations from ≈ 242°, lsr ≈ 60 km s −1 to ≈ 252°, lsr ≈ 40 km s −1 is completely missed by the lower sensitivity CO(1-0) map. Additional features can be identified, indicating the structure between the spiral arms towards this region of the outer Galaxy to be more complex than indicated by the data from Dame et al. (2001). We will discuss this in more detail in Section 5. Table 4. Velocity components identified from the CO(2-1) observations along 10 lines-of-sight. Source names starting with a 'G' are measured emission at targeted positions, whereas names starting with an 'O' indicate emission in the off-position. Uncertainties for lsr are in the order of 0.5 km s −1 , distance uncertainties in the order of 0.3 kpc.  Table 5. Clouds (i.e. coherent groups of velocity components) identified from the CO(2-1) observations along a given line-of-sight for 15 complexes. Source names starting with a 'G' are measured emission at targeted positions, whereas names starting with an 'O' indicate emission in the off-position.

Kinematic distances and Galactic distribution
We determined distances for all velocity components by applying the Brand & Blitz (1993) rotation curve, assuming a distance to the Galactic centre of R 0 = 8.34 kpc and an orbital velocity of θ 0 = 240 km s −1 , as derived by Reid et al. (2014) and used by Urquhart et al. (2018). We prefer the rotation curve from Brand & Blitz (1993) over the more recent one from Reid et al. (2014), as the latter is derived from parallactic distances to objects visible from the northern hemisphere and does not include measurements for the southern outer Galaxy. In Fig. 6 we show the distribution of our sources in the 3rd Galactic quadrant as viewed from the Galactic north pole. We find the closest source to be located at a heliocentric distance of R hel = 0.40 kpc and the most distant sources at R hel ∼ 12 kpc. This results in the sources spanning a range of Galactocentric distances between R gal = 8.50 kpc and R gal ∼ 16.5 kpc. Note that the distances are not only affected by the uncertainty of the velocity measurement, but also by the spread due to streaming motions ( 10 km s −1 ; Reid et al. 2014), the expanding supershell (∼7 km s −1 ; McClure-Griffiths et al. 2006), and the accuracy of the rotation curve, which can change the distance of any given source as determined from the rotation curve significantly (∼±1 kpc), effectively dominating the uncertainty of the distance estimates. In Fig. 7 we show the histogram of Galactocentric (upper panel) and heliocentric (lower panel) distances for all velocity components above the 5σ rms noise level. The peaks are not only tracing the spiral arms, but also result from complexes lying at similar Galactocentric distances.   This can also be seen in Fig. 5, where the 4 peaks are marked by the coloured dashed lines in the same way as in Fig. 7. We find that the peak located at r gal ∼ 9 kpc traces local gas including emission from the Vela molecular ridge. The peak at ∼11 kpc would trace the Perseus arm, if it was present in the area (see discussion in Sect. 5). The peaks around r gal ∼ 12.5 kpc (cyan dashed line) trace emission not associated with a spiral arm, being dominated by the huge complex at ∼242°at lsr ∼70 km s −1 (i.e. behind the supershell; see Section 5.2). The peak found at r gal ∼ 14 kpc partially arises from emission from the Outer arm as well as from structures between the Perseus and Outer arm located at ∼255°. Similarly, we see a number of peaks in the heliocentric distance histogram, but due to the projection these cannot be assigned to the arms, as the arms are spread out over a larger heliocentric distance range between 225°≤ ≤ 260°than with regard to Galactocentric distance.

Physical Properties
We are now going to investigate the physical properties that can be derived from archival mid-infrared to sub-millimetre continuum emission data in combination with the distances obtained from our pointed CO observations (Section 2.5). First we will discuss how we obtained the dust spectral energy distributions for the outer Galaxy. We will then present the derived physical properties and investigate how consistent they are, followed by a detailed look at the results. Here we will mainly investigate the star formation relations as well as the influence of the main structures present in the outer Galaxy on the physical properties of the observed clumps.

Dust spectral energy distributions
To obtain and fit the dust spectral energy distributions (SEDs) we follow the procedures we described in detail in König et al. (2017) and Urquhart et al. (2018), hence we only give a brief overview here. We use archival mid-infrared to sub-millimetre continuum maps to obtain the SEDs in 9 different bands at 8, 12, 14, 21/22, 70, 160, 250, 350 and 500 µm . In contrast to our previous work, there is no ATLASGAL data available at 870 µm for the outer Galaxy, so we use the 250 µm positions and source sizes obtained from the source extraction process as described in Section 2 for the photometry. In the far-infrared to submm regime we use Herschel This data is complemented by mid-infrared maps from WISE (Wright et al. 2010) and MSX (Price et al. 2001) covering the wavelength regime between 8 µm and 22 µm that is mostly dominated by the emission of a hot embedded component of (proto-)stars. The fluxes for each band are obtained through aperture photometry as described in detail in our previous work (Urquhart et al. 2018), reconstructing the SED of each source. Whenever a flux was measured in the WISE emission maps at the 12 or 22 µm bands, it was preferred over the corresponding fluxes at 12 or 21 µm determined for the MSX bands due to the lower noise level and moderately better resolution of the WISE images. In this way we obtained SEDs for all observed positions as well as for off-positions that were matched with the extracted sources. We have fitted the SEDs using a single component greybody or two-component model, depending on the emission in the mid-infrared bands. A sample two-component SED is shown in Fig. 8. In contrast to our previous work we use the emission found in the 350 µm SPIRE band as the reference wavelength due to the absence of a flux measurement at 870 µm. In total we were able to fit 611 SEDs (∼77%) of the 791 sources. The SEDs for 180 sources were either not recovered completely due to sensitivity, the source being located in a crowded region, the SED being irregular or the source was too close to the edge of the Hi-GAL area (or a combination of those). We summarize the pa- rameters used to obtain the SEDs and the fitted parameters (dust temperature and opacity) in Table 6 alongside the evolutionary classes as determined from the SEDs (see Section 4.1).

Deriving physical properties
Using the dust temperatures from the fitted SEDs and assigning distances as determined in Section 2.5, we calculate the physical properties from the SEDs as described in König et al. (2017). However, as the reference wavelength has changed to the S 350 SPIRE band, slight changes were made on how we calculate the clump mass and H 2 column density. We obtain the clump mass M clump from the integrated 350 µm flux density S 350 as where d is the distance to the source, B 350 (T d ) the intensity of a blackbody at 350 µm at the cold dust envelope temperature T c . The dust opacity κ 350 = 1.1 cm 2 g −1 at 350 µm is calculated as the mean of all dust models from Ossenkopf & Henning (1994), using the dust emissivity index of β = 1.75 used for fitting all Table 7. Physical parameters derived from the dust SEDs for the first 15 sources: bolometric luminosity L, clump mass M, peak column density N H 2 , and luminosity-to-mass ratio L/M. Full table available online at CDS.

Consistency checks
To check the consistency of our method, we compare the N H 2 column densities derived from the SEDs with column densities derived from the 12 CO(2-1) emission. To obtain the column densities from 12 CO(2-1), we calculate them as where we use the H 2 -to-CO conversion factor of X12 CO(2−1) = 2.3 × 10 20 cm −2 (km s −1 ) −1 obtained by Brand & Wouterloot (1995) for the outer Galaxy, and the integrated line intensity I( 12 CO(2 − 1)) as measured from the observed spectra with a line ratio of 12 CO(2-1)/ 12 CO(1-0)= 0.7 (Sandstrom et al. 2013).
In the upper panel of Fig. 9 we show the comparison of the column densities derived from dust and CO. As the CO(2-1) column density is derived from emission measured within a beam of 30 , it traces the denser regions of the more extended clumps with an average angular source size of 62 , resulting in slightly higher column densities when compared to those derived from the dust emission. Nevertheless, we find both quantities to be in good agreement (p-value of 0.0155) although a large scatter is observed. We also compare our results to a similar sample of southern outer Galaxy sources from Elia et al. (2013) obtained for an adjacent area of the southern sky (216.5°< < 225.5°).
In the lower panel of Fig. 9 we compare the clump masses as derived from the Herschel dust continuum emission of both samples. Similar to the masses calculated for our sample, we apply a correction factor for the varying gas-to-dust ratio found by Giannetti et al. (2017a) to the masses calculated by Elia et al. (2013). We find the distribution to be statistically different (pvalue of 5.3 × 10 −4 ) with the sample of the present work picking up significantly more lower mass sources. This is reflected by the mean values to be almost identical with values of 58.2 ± 6.0 and 56.8 ± 5.4 M for our sample and the Elia et al. (2013) sample, respectively, but the median values to differ by a factor of 1.6 (10.1 and 16.2 M , respectively). The difference for these two samples is likely caused by a combination of two effects. First, the areas are non-overlapping, so the differences might reflect intrinsic differences of the two areas covered in the outer Galaxy, especially when taking into account the Galactic supershell GSH242-03+37 covered in the present work (see Section 5.2). Furthermore, the distances for the sample by Elia et al. (2013) were obtained with the NANTEN 4 m telescope with a beamsize of 2.6 ( Kim et al. 2004), only allowing to assign distances to the brighter sources within the beam. From the comparison with the column densities derived from 12 CO as well as with the clump masses from Elia et al. (2013), we conclude that our methods can be considered reliable, as either the distributions are similar as shown by an Anderson-Darling test or agree on average within the margin of error.

Distance biases
Some physical properties suffer from observational distance biases, such as the bolometric luminosity, clump mass and linear source size. These biases are caused by the fixed sensitivity and resolution of the telescope/instrument used. Due to the limited sensitivity, only the brightest and most massive sources can be observed at the farthest distances. Similarly, the limited angular resolution allows us only to observe sources down to this apparent size, which at farther distances translates to larger linear source sizes. To avoid misleading trends that are introduced by these sensitivity and resolution based selection biases, we determine a completeness limit, above which the survey does not suffer from these selection effect out to the given distance. First we need to distinguish between distance independent parameters and distance dependent parameters. The dust temperature, peak column density and L/M are distance independent, as they are either intrinsic to the sources or cancel out the distance dependence. As mentioned in the last paragraph, the bolometric luminosity and clump mass are directly scaled by the distance squared, and hence are highly dependent on correct distances and are prone to distance dependent observational biases. The same is true for the linear source size, which is linearly dependent on the distance. To determine in which mass and luminosity range our survey is not suffering from distance biases, we show the distribution of luminosities, masses and linear source sizes with respect to heliocentric distance in Fig. 10. We calculate the theoretical sensitivity according to Equations 1 and 2 in König et al. (2017) for the bolometric luminosity and clump mass from , and linear source size (lower left) versus Heliocentric distance. The solid lines mark the distance dependent sensitivity/resolution limit. The horizontally dashed green lines mark the limit above which our survey is not suffering from a distance bias up to 9 kpc (vertical dotted line).
the minimum values as input parameters, varying the distance. Similarly we calculate the resolution limit from Equation 3 in König et al. (2017) but use the beam size for SPIRE 250 (18.2 ) as input parameter. The theoretical sensitivity and resolution limits are plotted as green solid lines in Fig. 10. As the source count of our sample drops significantly beyond 9 kpc heliocentric distance we estimate the completeness limit up to this distance as the value of the theoretical sensitivity or resolution limit at 9 kpc. For the bolometric luminosity, we find that sources out to ∼9 kpc distance are strongly affected by the sensitivity of this survey for luminosities below 6.5 L . Masses of 9 M are found to be the completeness limit for clump mass, whereas for the linear source size, sources smaller than 0.23 pc suffer from the distance bias.
We will filter our samples according to these completeness limits when analysing the distance dependent physical properties.

Discussion
In this section we will discuss the physical properties derived in the previous section. We put a focus on the analysis of the properties found in the outer Galaxy with respect to the structures found here.

Evolutionary Sequence
Based on the flux densities obtained for the SEDs we determine the evolutionary state of the sources, following the classification scheme introduced in König et al. (2017), which was also used by Urquhart et al. (2018) for the entire ATLASGAL sample. We follow here the naming scheme of our latest paper and refer to mid-infrared bright sources as young-stellar-objects (YSOs). Note however, that these might also include some compact Hii regions, which we did not identify individually. We will therefore refer to the following three classes: -Quiescent sources, which are dark at 70 µm (no compact emission at 70 µm) and represent the earliest phase of starformation in our sample. These sources might or might not be collapsing. -Protostellar sources, which are bright at far-infrared and submm wavelengths but are not sufficiently evolved to produce significant emission at mid-infrared wavelengths (F 20 ≤ 0.1 Jy) 3 . These sources are in the process of collapse and are internally heated. -Young-stellar-objects (YSOs) which are bright at midinfrared wavelengths (F 20 > 0.1 Jy). This group of sources is significantly evolved to produce strong emission at midinfrared wavelengths by an internal heating source.
This classification scheme has been proven to be reliable as shown by Giannetti et al. (2017b) using molecular line data. It was further bolstered by our recent work on the ATLASGAL sample (Urquhart et al. 2018), showing clear trends for increasing dust temperature and the bolometric luminosity to clump mass ratio, proving this evolutionary sequence. Cumulative histograms for the dust properties and the three evolutionary phases of all sources are shown in Fig. 11. Again, we find the evolutionary stages to be well separated from each other for the dust temperature, the bolometric luminosity and the luminosity-to-mass ratio, as found in König et al. (2017) for the Top100 sample 4 . This reflects the increase in luminosity due to the accretion of material onto the protocluster and the heating of the local environment as the protostars evolve (compare Sect. 4.3 and Molinari et al. 2008). The similarity between the three evolutionary classes is determined by Anderson-Darling tests, indicating the distributions to be significantly different for each of the aforementioned physical properties, as their pairwise p-values are all well below the threshold of p < 0.0013, corresponding to a 3σ confidence. These three properties (dust temperature, bolometric luminosity and luminosity-to-mass ratio) are therefore excellent indicators of the evolutionary phase of a clump. Care has to be taken though, as there is a large overlap between the different phases, and therefore we can not simply read of the evolutionary phase of a single clump by only taking into account these numbers. However, they can be used to determine statistical properties and identify significant trends in the data. For the remaining three physical properties (clump mass, linear source size and H 2 column density), the quiescent and protostellar evolutionary stages are indistinguishable with a p-value of 0.1. Only the clumps in the YSO phase show, on average, slightly higher values of these properties than the two earlier phases with p-values 3 F 20 is the flux in the 21 µm or 22 µm band from either MSX or WISE. 4 The ATLASGAL Top100 sample (König et al. 2017) defines a sample of the approx100 brightest sources in 4 distinct evolutionary stages selected from the ATLASGAL survey.  below 1.8 × 10 −4 . This indicates that the classification scheme tends to identify the larger, more massive clumps as being more evolved, indicating that the more massive clumps evolve significantly faster than their low-mass counterparts.

Physical properties
Temperatures and optical depth: Temperature and optical depth are determined as fit parameters from the SEDs and are distance independent. We find that the dust temperatures range from 9.65 K to 41.31 K with a mean value of 17.29 K. We point out that our sample has on average 2 K lower mean temperature when compared to Urquhart et al. (2018), probably arising from the difference in sensitivity of the two surveys. We will therefore leave a detailed comparison for Paper II. For the optical depth we find that all of our sources are optically thin (τ 1) at 350 µm, as the optical depths range from 1.9 × 10 −6 up to a maximum of 2.9 × 10 −3 .
Source sizes: In Fig. 11, lower-left panel, we show the distribution of linear source sizes for our sample. Linear source sizes derived from the apparent full-width at half maximum source size vary between 0.03 pc and 1.78 pc yielding a mean source size of 0.25 pc. This is almost identical to the value of 3.2 × 10 −1 pc for the ATLASGAL sample in the 1st and 4th Galactic quadrant (Urquhart et al. 2018), showing that both samples trace structures of similar scale, allowing for a detailed comparison of the different samples.
H 2 Column Density: The peak H 2 column density is found to range between 6.3 × 10 19 cm −2 and 4.1 × 10 22 cm −2 with a mean value of 3.8 × 10 21 cm −2 in the outer Galaxy. As this is below the 10 23 cm −2 found by Urquhart et al. (2018) above which almost all clumps are associated with high mass star formation, we expect to find significantly less massive clumps in the outer Galaxy.
Bolometric luminosity and clump mass: The bolometric luminosities for the outer Galaxy sample range between 8.0×10 −2 L and 2.4×10 4 L with a mean value of 3.6×10 2 L . Clump masses range from 2.3 × 10 −2 M to 1.7 × 10 3 M with a mean value of 58 M . As with the H 2 column density, we expect from these maximum values only a small number of sources to be able to form high mass stars, which we will investigate in the following sections.

Star formation relations
In Fig. 12 we show the bolometric luminosity plotted versus the clump mass. For a given clump one expects the luminosity to increase during the accretion phase, hence moving upwards in the figure until stars reach the zero age main sequence (diagonal solid line). As soon as stars are formed, their feedback disrupts a given clump. As a consequence the clump gets dispersed, decreasing its mass, hence moving left in the diagram. These evolutionary paths through the diagram are indicated by the grey tracks, as calculated by Molinari et al. (2008). In order to visualize the differences between the evolutionary phases as described in Section 4.1, we show the linear fits to the evolutionary classes as dash-dotted lines. As can be seen, the sources of the different classes move up in the plot as expected from the evolutionary scheme described in Section 4.1. This trend is consistent with the cumulative distribution of the bolometric luminosity-to-clump mass ratio L /M shown in the lower right panel of Fig. 11. There we find the three classes to be well separated with all pvalues well below 0.0013, following the expected evolutionary trend. Further we indicate the expected luminosity for a B2 star (∼ 8 M ) as calculated by Mottram et al. (2011) as dashed horizontal dashed line. Above this threshold we assume that at least one massive star has formed in the clump being responsible for most of its luminosity, but point out that this threshold only sets an upper limit for the number of high mass stars according to our criterion. The vertical dashed line marks the clump mass threshold above which it is likely that the clumps host massive dense cores or a high-mass protostar (Csengeri et al. 2014). From these thresholds we see that only the minority of the sources in the outer Galaxy are able to form a high-mass star (32 sources) and the majority of these have already done so (24 sources at most).

Mass-size relation
In Fig. 13 we show the clump mass plotted versus the source radius. The grey dotted line marks the mass-limit for high-mass stars at 8 M . The grey shaded area marks the regime where supposedly only low-mass stars form as determined by Kauffmann et al. (2010), with only 40 sources (7%) of the outer Galaxy sample above the threshold and the mass limit. The black dashed line marks the lower limit for effective high-mass star formation as determined by Urquhart et al. (2014c) indicating a surface density threshold of 0.05 g cm −2 found for the inner Galaxy using the ATLASGAL survey. 140 of our sources (i.e. 23%) are found above this threshold and the mass-limit for high-mass stars. Although we find the majority of the clumps to be able to form only low-mass stars, as we did in the last section, we find a significantly higher fraction to potentially form high-mass stars according to the latter threshold used here. As a consequence we speculate that the threshold for high-mass star formation based on mass surface density is likely to be different in the inner and outer Galaxy, as the radial mass surface density profile of the Galaxy drops significantly according to the review by Heyer & Dame (2015; Fig. 7). Finally, we compare the evolutionary phases (colour-coded), but no trend with regard to the evolutionary state of the clumps is found, as can be seen from the linear fits to the individual classes (coloured dash-dotted lines). This is in agreement with our findings in Sect. 4.1.

Independent high-mass star formation tracers
To further identify high-mass star forming clumps, we have also investigated the presence of methanol Class ii masers as identified by the Methanol MultiBeam survey (MMB; Green et al. 2012), which are thought to be exclusively associated with objects in the early phases of high-mass star formation . In the whole area of our survey, only 2 methanol Class ii masers are found (G254.880+0.451 and G259.939-0.041), indicating either a lower rate of high-mass star formation, or being the result of the lower metallicity towards the outer Galaxy (Lemasle et al. 2018), and thus reducing the available amount of methanol CH 3 OH, which contains two metal atoms. Therefore we also investigated the number of Hii regions in the survey area as identified by the Red MSX Survey (Lumsden et al. 2013) and the WISE catalogue of Galactic Hii regions (Anderson et al. 2014), as the formation of Hii regions is independent of the metallicity. Merging both catalogues we find 10 known Hii regions in our survey area (3 from the RMS survey, 8 from the WISE catalogue with one source in both), as well as another 24 candidate Hii regions from the WISE catalogue. These are about a factor 20-30 less sources per unit-area as compared to the inner Milky Way for the aforementioned surveys. As these numbers are in the same order as the results from our dust continuum emission in the previous paragraphs, we conclude that there is only very limited high-mass star formation occurring in the outer Galaxy.

Structure of the outer Galaxy
In this section we use the distance information as derived from the radial velocities to give an overview of the structure of the area investigated in this work. We will briefly investigate the spiral structure as found toward the 3rd Galactic quadrant, and look into the influence of a Galactic supershell known to exist in the observed region.

Spiral arms and inter-arm regions
The nature of the structure of the Milky Way is still under debate (Dobbs & Baba 2014), and it is unclear weather it is a 'flocculent' or 'grand-design' spiral galaxy. The existence of four spiral arms would denominate it as a flocculent galaxy, whereas an interpretation as grand-design would leave the Galaxy with two density wave arms (Perseus and Scutum-Centaurus) and the others being transient features. For this reason, the location and visibility of the spiral arms towards the observed Galactic quadrant is also still under debate with either a single arm or two arms visible (compare e.g. Reid et al. 2014;, Koo et al. 2017. Here we will focus on a model where both, the Perseus arm and the Outer arm, are found in the southern outer Galaxy, as the Outer arm is clearly present in the Hi emission (compare Fig. 14). However, the location of the arms in the 3rd Galactic quadrant is not well constrained, as their loci are extrapolated from the northern hemisphere, where the measurements of spiral arm positions are better constrained. We show a larger region towards the outer Galaxy in Fig. 14 as described in Reid et al. (2014): where R gal is the Galactocentric radius, R ref the Galactocentric radius at the reference Galactocentric azimuth β ref (with β ref = 0 towards the sun, increasing with Galactic longitude), and the pitch angle Ψ. In Fig. 14 we indicate the extrapolation of the spiral arms and Orion/Local spur as determined by Reid et al. (2014) and Reid et al. (2019) from trigonometric parallax measurements for the northern hemisphere as dashed lines. This includes (from low to high lsr ) the Orion/Local spur, the Perseus arm and the Outer arm. Furthermore, we plot the two arms visible in the 3rd Galactic quadrant from the 4-arm model as deter-  (2015) as dash-dotted lines. Although the extrapolations for the Perseus spiral arm are all still in good agreement, the extrapolations for the Outer arm deviate quiet drastically from each other (esp. Reid et al. 2019) due to the high uncertainties involved, and fail to trace the highest column density Hi emissions in the present region. Therefore we also plot our own estimates as solid lines for the 3 main features (from low lsr to high lsr ): local emission from the Vela molecular ridge (lower left) to the Orion/Local spur (lower right), the Perseus arm (middle) and the Outer arm (upper). As a detailed analysis of the position of the spiral arms is out of the scope of this work, we have manually modified the spiral arm parameters such as to visually match the densest emission as seen in the Hi integrated intensity map. Note that the local emission (Vela molecular ridge and Orion/Local spur) is an interarm feature between the Sagitarius spiral arm located in the inner Galaxy at R gal ∼ 7 kpc and = 0°(not visible here) and the outer Galaxy. The position indicated here by the solid cyan line therefore does not represent an arm that is part of the spiral pattern but just a fit to the local emission. In case of the Vela molecular ridge the emission distribution bends inwards with increasing Galactic azimuth with a large negative pitch angle, in contrast to the spiral arms extending outwards with increasing Galactic azimuth and a comparably smaller pitch angle. The parameters used for the spiral arms and the local emission are summarized in Table 9. To give an overview and explore the three dimensional structure of the observed region, we present longitude-velocity ( − v) plots for different slices of Galactic latitude in Fig. 15 and latitude-velocity (b − v) plots for different slices of Galactic longitude in Figs. 16 and 17. We show the Hi integrated intensity as a background image (McClure-Griffiths et al. 2009), the CO(1-0) emission from Dame et al. (2001) as black contours, and mark the positions of our sources observed in CO(2-1) as small crosses with those associated with dust clumps as coloured dots. We will now discuss the different features present in the area. We find that our sources as detected in 12 CO(2-1) are mostly well correlated with the brightest features of the Hi emission. They are tracing the local emission very well, but except for the region of the brightest Hi emission located at ∼ 233°, are poorly tracing the Perseus arm. We conclude that this is not a sensitivity issue, as plenty of sources between lsr ∼80-90 km s −1 are well detected in the ∼ 258°region, and therefore we find no reason why such sources should not be detected in the Perseus arm be-tween 225°< < 250°. Similarly, we find the CO(2-1) emission to be poorly correlated with the Perseus arm for some slices of longitude ( ∼ 254°and ∼ 227°, Fig. 16, right panels). We rather conclude that the poor coherence of the observed source velocities with the locus of the Perseus arm in − v and b − v plots is a consequence of the presence of the expanding Galactic supershell G 242-03+37, which we will investigate further in Sect. 5.2. Several larger features can be identified spanning between the local emission and the suspected positions of the spiral arms (compare Fig. 15): In the region towards higher longitudes than the supershell (250°< < 260°) a web-like network of blobs, bridges or spurs can be seen, with large clusters of clumps correlated with the brightest Hi emission. At least four such complexes can be identified by eye from the − v maps ( ∼ 254°, lsr ∼ 38 km s −1 ; ∼ 257.5°, lsr ∼ 50 km s −1 ; ∼ 258°, lsr ∼ 80 km s −1 ; ∼ 259°, lsr ∼ 60 km s −1 ), with more isolated clumps along the bridges. These bridges are also well visible in the b−v plot around ∼ 285.5° (Fig. 16, upper right panel). Note that these bridges might not be visible in CO(1-0) the (black contours Dame et al. 2001), as the sensitivity is rather poor (rms noise up to 0.4 K). Between 225°< < 240°(i.e. lower longitudes than the supershell) we find at the lower latitudes (see right hand column of Fig. 15) a feature in Hi that is aligned perpendicular to the line-of-sight, indicating the continuation of the Perseus arm from the north. But for higher latitudes (left hand column of Fig. 15) this structure becomes more complex with a network of bridges spanning between the Orion/Local spur and the Outer arm more parallel to the line-of-sight (and perpendicular to the arms). This can also be seen in the b − v plot in Fig. 16, lower left panel, where the CO emission at b ∼ 0°is spanning from lsr ∼ 20 km s −1 out to lsr ∼ 60 km s −1 , whereas the local emission and the Perseus and Outer arm are more clearly separated at lower latitudes. This change in morphology and orientation in dependence on the Galactic latitude could be either interpreted as the Perseus arm being disrupted or this structure not being a spiral arm at all. In contrast to the web of structures located towards higher longitudes than the supershell between 250°and 260°, we only find one larger star forming complex in this area ( ∼ 232°, lsr ∼ 50 km s −1 ), located at the rim of the supershell (Fig. 17). In general, we find the three-dimensional structure of the Galactic disk towards the outer Galaxy in the third quadrant to be rather complex; not only with respect to longitude and velocity/distance, but also with respect to the vertical structure of the thin disc. The structures found in the observed region change rather dramatically with Galactic latitude, showing that the thin disk is a complex three-dimensional web-like structure than a flat pan-cake-like structure.

The Galactic supershell GSH 242-03+37
One of the most striking features towards the observed region is the void located at ∼ 242°and lsr ∼ 40 km s −1 . This structure is known to be the Galactic supershell GSH 242-03+37 and was first identified by Heiles (1979) and more recently investigated by McClure-Griffiths et al. (2006). It has a diameter of the order of a kiloparsec and an expansion velocity of ∼ 7 km s −1 . With the shell spanning about 15°in Galactic longitude, 25 km s −1 in velocity and centred −1.6°below the Galactic plane, we find that our observed region spans roughly a band around the equator of that supershell. As this structure has a profound impact on this region, we will briefly discuss the possible origin of this supershell. McClure-Griffiths et al. (2006) calculate the energy needed to form such an expanding supershell to be in the order of 10 53 ergs, which is about two orders of magnitude higher than the expected energy input from a single Type ii supernova (Woltjer 1974). This would therefore require in the order of hundreds of Type ii supernovae to drive this shell, which is rather unlikely, especially as there is no strong X-ray source at the centre of the shell, which would be expected in the presence of hundreds of expected supernova remnants. A more suitable explanation might be the passage of a high velocity cloud (HVC) through the Galactic disk, like the one observed by Park et al. (2016) causing the Galactic shell GS040.2+00.6-70 in the northern hemisphere. Furthermore, HVCs are known to punch holes through galactic discs (e.g. Schulman 1996, Boomsma et al. 2008. Preliminary literature research further bolsters this hypothesis (compare e.g. Galyardt & Shelton 2016, McClure-Griffiths et al. 2006, Planck Collaboration et al. 2015, but as an investigation of the origin of this shell is out of the scope of this work, we will instead con-  Fig. 19. Histograms of the sensitivity filtered sources (see text) for the distance to the centre of the supershell (left) and Heliocentric distance of this subsample. The blue shaded area marks the distance range spanned by the shell.
tinue to discuss its impact on the region. A consequence of the existence of this huge expanding supershell is a reduced reliability of kinematic distances for sources within its vicinity, as the expansion of the shell at a velocity of v exp ∼ 7 km s −1 (McClure-Griffiths et al. 2006) distorts the Galactic rotation curve. For the distances to the shells center this mostly affects the sources at the front and rear wall of the supershell, as the line-of-sight velocity lsr is parallel to the expansion velocity v exp . However, the expansion velocity v exp has no effect on the sources on the northern or southern edges of the shell, as their position relative to the center of the shell is measured from the longitude and the line-of-sight velocity lsr is not affected as it is perpendicular to the expansion velocity. The effect can be seen in Fig. 6 as the supposedly circular supershell is deformed into an ellipse along the line-of-sight. Accordingly, this needs to be taken into account when interpreting structures and especially spiral arms in this region of the Milky Way. Furthermore, the supershell is coincident with the extrapolation of the Perseus arm, which might be the main reason why the Perseus arm cannot be traced clearly in this part of the outer Galaxy. If a large high-velocity cloud has hit the Perseus arm in the past, such an event might simply have lead to its local destruction, pushing the material of the arm away. Indeed, we see a ring of sources identified in CO emission around this supershell. In Fig. 18 we present a histogram of the number of CO emission components, clouds and associated dust clumps vs. the centre of the supershell normalized by the area. With the Galactic plane being coincident with the rim around the equator GSH 242-03+37, we find the central region (r GSH ≤ 0.4 kpc) void of any clumps. However, we find that the number-density of sources is increased in the wall of this supershell between 0.6 kpc≤ r GSH ≤1.2 kpc, which we estimate to be ∼0.6 kpc wide. In fact, we find a sharp increase in number-density of clumps from the inner void to the wall, which then gradually falls off again 5 . This is in agreement with McClure-Griffiths et al. (2006), finding a sharp increase in Hi from the inner region to the wall of the shell, interpreting it as an indication of compression and being associated with a shock.

Physical Properties with respect to the Galactic Supershell
In the previous section we had a look at the distribution of clumps around the supershell. In this section we will take a look at the impact of the supershell on its environment. To avoid observational and selection biases, we limit our sample to only include sources that are symmetrically distributed around the shell with a maximum distance of 2.5 kpc to its centre, reducing our sample size to 429 sources. Furthermore, we need to filter our sample for sensitivity and resolution thresholds, with the farthest source in this sub-sample is located at 6.2 kpc Heliocentric distance. For this distance we limit our sample to sources with L bol > 2 L , M clump > 3 M , and r src > 0.15 pc according to Fig. 10, reducing our sample to 178 sources. In Fig. 19 we show two histograms for this subsample. First we show the distribution of sources per unit area with respect to the distance to the shells centre (left). Again we are able to identify the central void of the supershell extending up to ∼ 0.5 kpc as well as the enhanced number of sources located in the shells wall. In the right panel, we show the histogram of heliocentric distances, with the range covered by the supershell marked as blue shaded area. We find no bias to any heliocentric distance, although peaks from local clusters are clearly visible. These are either from lo-terial in the walls of the supershell might be shocked and compressed. Furthermore, the increased source count per unit area fits perfectly into the picture of Izumi et al. (2014), who found star formation to be possibly induced by the passage of a highvelocity cloud through the disc in the outer Galaxy in the 2nd Quadrant.

Summary and Outlook
In order to extend our previous studies of the ISM and star formation (König et al. 2017, Urquhart et al. 2018 to the outer Galaxy between 225°≤ ≤ 260°we used Herschel/Hi-GAL 250 µm SPIRE continuum emission maps to select a representative sample of more than 800 sources from a rudimentary source catalogue of more than 25,000 extracted clumps using SExtracor (Bertin & Arnouts 1996), giving positions and source sizes for these clumps. We observed these sources in 12 CO(2-1), identifying 1248 clouds that consist of a total of 1383 individual velocity components, for a total of 1090 positions, including recovered off-positions. 966 (88.6%) lines-of-sight were found with a single velocity component, whereas two or more clouds were found towards 87 (11.4%) lines-of sight, yielding on average 1.2 clouds per line of sight. Consecutively, distances were calculated using a rotation model of the Galaxy, applying the rotation curve from Brand & Blitz (1993) for all clouds and velocity components. For every line-of-sight, we finally associated the cloud with the highest integrated intensity to the according dust clump. Combining our velocity measurements with Hi emission maps from the GASS survey (McClure-Griffiths et al. 2004) and CO(1-0) maps from Dame et al. (2001), we were able investigate the large-scale structures of the southern outer Galaxy between 225°≤ ≤ 260°. We determined physical properties of the selected dust clumps, recovering their dust spectral energy distributions from Hi-GAL, MSX and WISE continuum emission maps. The SEDs were consecutively fitted with a simple two-component model yielding dust temperatures, integrated fluxes and H 2 column densities. Combining the results with the kinematic distances determined from the CO emission allows us to calculate physical properties such as bolometric luminosities and clump masses. To guarantee the consistency of our data with other studies, we compare the peak H 2 column densities obtained from dust continuum emission against the column density derived from the 12 CO(2-1) emission. Although deviations of up to an order of magnitude are found for individual clumps, we find a good agreement for the general trend, and allot the deviations to local variations in the gas-to-dust ratio and CO-to-H 2 conversion factor. Furthermore, we compare the clump masses of our sample to a similar sample from (Elia et al. 2013), finding the mean values to be almost identical, showing our methods to be reliable. Our main findings are: (i) In general, we find the positions of the identified CO clouds to be strongly correlated with the highest column density parts of the Hi emission. On the other hand, we were also able to identify a web of bridges, spurs and blobs of star forming regions spanning between the larger star forming regions, unveiling the complex three-dimensional structure of the outer Galaxy in unprecedented detail. Although the latter might be an indication of the outer Galaxy to be of a flucculent nature, a definite answer is difficult due to the influence of a large, expanding supershell (GSH242-03+37) in the survey area.
(ii) For the investigated clumps, we find the evolutionary stages to be well separated by the dust temperature, bolometric luminosity and luminosity-to-mass ratio, consistent with our results in (König et al. 2017). However, we find the clump masses and peak column densities to be similar in the starless and protostellar phase, but find these quantities to be significantly higher for the YSO phase, indicating that the more massive clumps evolve significantly faster than their lower mass counterparts.
(iii) For the outer Galaxy we find only 24 sources with an inferred luminosity higher than that of an early B-Type star (∼8 M Mottram et al. 2011), the masses of only 8 sources above the threshold where it would be likely to host a massive dense core or high-mass protostar according to Csengeri et al. (2014) or at most 40 sources above the threshold for high-mass star formation as determined by (Kauffmann et al. 2010). Even more so, we only find 2 methanol Class ii masers, and 10 known as well as 24 candidate Hii regions in the whole survey area of the outer Galaxy, indicating only a low fraction of star forming regions in the outer Galaxy to be able to form high-mass stars.
(iv) Investigating the influence of the expanding Galactic supershell GSH 242-03+37 in detail, we find the physical properties and star formation activity of sources located within the walls to be not statistically different from sources located farther away. Nevertheless, as we have seen in Section 5.3 we find the number-density of sources increased within the walls of the supershell, leading us to the conclusion that the expanding supershell supports the formation of clumps, but once they collapse has no further influence.
In Paper II we will use the sample of star forming regions characterized in the present work in order to compare the properties of the outer Galaxy with those found for the inner Galaxy and derive global trends for the Milky Way.