SPECTROSCOPIC BINARIES IN THE ORION NEBULA CLUSTER AND NGC 2264

M2FS is a multi-object spectrograph on the Magellan Clay Telescope that is capable of both low-resolution and high-resolution echelle spectroscopy. Up to 256 targets can be observed over a 29' field of view. The fibers observing these targets are split into two independent yet identical spectrographs. M2FS fibers need to be plugged in manually into predrilled plates. The minimum allowed separations between fibers is 12''. A slitwidth of 180 μm yields a typical resolution of R ∼ 20,000.

We observed a total of four fields toward the ONC and two fields toward NGC 2264 in 2013 November and 2014 February (Table 1) using M2FS. We targeted a subset of stars from the T09 and T15 samples that was previously thought to be varying and/or had multiple reliable detections. The Mg i filter was used covering a  wavelength range of ∼5100–5210 Å.

In addition, we also observed three fields toward the ONC in 2014 December with M2FS, but using the Hα and Li i filters covering range of 6525–6750 Å. Since two separate orders are observed simultaneously, only 128 targets can be observed in this configuration. We mainly obtained spectra for those sources that were originally identified by T09 as blueshifted relative to the gas in order to confirm their membership in the cluster via the presence of the Li i line, which can be used as an indicator of youth (Briceno et al. 1997). Additionally, given the available fibers, we observed objects that were previously monitored and had V fluxes between 12 and 13.5 mag.

Data were reduced using a custom Python code written by J. Bailey to merge the data and subtract the bias, and the IRAF pipeline HYDRA to trace the orders, extract the spectra, calculate and apply the wavelength solution using a set of Th–Ar exposures, and perform sky subtraction.

Spectra taken with the Li i and Hα filters were particularly affected by the strong nebular emission lines from S ii (6717 and 6731 Å) and N ii (6549 and 6583 Å), as well as Hα. These features are always narrow and appear in conjunction with each other in any given spectrum. They would be present as the only features even in the spectra of stars that were too faint to be detected. All of these features were masked out in the final data product if they were observed during the visual examination. However, because the Hα nebular line often interfered with the line that should have been observed due to stellar emission, often superimposed near the center of the line or barely offset from it, the masking process makes it impossible for us to reliably measure the equivalent widths of Hα for most spectra, and prevents us from detecting narrow stellar lines (both emission and absorption) in nearly all of the sources.

In addition to these narrow features, some spectra exhibited very broad and strong emission-like features at 6600 and 6725 Å in Hα and Li i, respectively, spanning ∼20 Å in width. These features appeared in approximately the same pixel range in both orders and are thought to be caused by Littrow ghosts from the optics. In the data taken with the Mg filter, a narrower and weaker feature appeared at 5181 Å in the "blue" spectrograph, and at 5187 Å in the "red" one. They are expected to have similar origins. In those cases where these features appeared to be significant, they were masked out.

In addition to the T09 observations of the ONC made in 2007 with Hectochelle, data were also acquired in 2008 and 2009 (Table 1). The multi-fiber echelle spectrograph Hectochelle on the MMT has a 1° field of view and can observe up to 240 targets simultaneously which can be positioned via robotic arms. The RV31 filter was used to cover the wavelength range ∼5150–5300 Å with a typical resolution of R ∼ 35,000. The data have been reduced using an IRAF pipeline developed by G. Fűrész. A more detailed description of the Hectochelle data reduction can be found in Sicilia-Aguilar et al. (2006).

The stars that were previously observed by T09 in ONC using MIKE fibers on the Magellan Clay telescope had an additional epoch observed in 2008 November (Table 1). MIKE consists of two independent spectrographs that can observe 128 fibers each. One of the spectrographs was used to cover the wavelength ∼5120–5190 Å, and the other one to cover ∼5140–5210 Å, at a resolution of R ∼ 18,000. A description of the data reduction is available in T09.

Out of the 137 sources originally identified by T09 as RV variables in the ONC region, we can confirm only 15 as such in our final catalog. The remaining sources exhibited either little to no change in velocity or had variable velocity measurements that were of low significance. Some of those sources could still be multiple systems, but we do not consider them further here due to our stricter limits for significance to avoid false positives. Upon closer examination, those sources that were previously identified as double-lined binaries by the presence of the second peak in the correlation either could not be confirmed as such or were flagged as variable by our method. For this reason, we focus only on the 130 sources that we can identify as only RV variables with either new or reanalyzed spectra.

We consider two possible causes of the RV variability. First, it could be due to the orbital motion of the muliple systems. Second, the changing RV could be a result of RV jitter due to spots on the surface of magnetically active stars. The typical effect of jitter in main-sequence stars is on the order of a few m s⁻¹ (Hillenbrand et al. 2015), but it could be on the order of ∼1 km s⁻¹ in pre-main-sequence stars (Donati et al. 2013, 2014, 2015).

Given that RV variable sources typically had significant fluctuation from v_ave, RV jitter can only account for a handful of sources. We identify sources as multiple if they have at least one measurement $| {v}_{\mathrm{ave}}-v| \gt 4$ km s⁻¹. A total of 113 sources satisfy this requirment. The remaining 17 sources fall below this threshold; while they do exhibit RV variability, we cannot confirm that it is due to orbital motion within a multiple system. Because of this, we exlcude them from any calculations involving the multiplicity fraction.

The velocity curves for most of our sources are very undersampled, and so we could fail to detect real RV variables. We consider the minimum number of epochs required for a guaranteed detection of an RV varying system to be three. With only 2 measurements, it is possible to miss a binary system due to coincidental timing between the observations. However, with three or more epochs of data, the fraction of the number of systems identified as multiples to the total number of stars remains relatively unchanged (Figure 1).

Zoom In Zoom Out Reset image size

We attempted to identify false positive sources that were flagged as multiples if they had only a single discrepent measurement from the mean. To do that, we calculated the reduced χ² while thowing out the single most variable RV measure from v_ave and identified sources where this revised reduced χ² < 16. We required that (a) all of the sources flagged by this method have at least four measurements with R > 6, since otherwise the lack of detected variability could be due to poor sampling; (b) at least four epochs during which the measurements were taked are separated in time by more than two days and, as such, short separations in time cannot detect variability due to orbital motion with orbits longer than a few days; and (c) only a single measurement has a discrepent RV, including R < 6 measurements if they have comparable variability from v_ave. This identified six sources: RV 10, 138, 929, 1372, 1496, and 1660. In the cases of RV 1372, 1496, and 1660, we provide detailed orbital fits (Section 4.3). These fits allow little doubt as to the nature of these sources as muliple systems; however, without a single strongly deviating RV they would not have been identified as such based on the data in this paper. For this reason, while the remaining three sources (RV 10, 138, and 929) are flagged in Table 3 as possible false positives, we include them in the flollowing calculations.

The observed multiplicity fraction (defined as the overall number of the multiple systems, hereafter MF) within the RV data set is 8.0 ± 1.2% toward the ONC and 6.7 ± 1.1 toward NGC 2264 if we include all of the sources in the Tables 2 and 3. However, if we require v_ave, converted from the heliocentric to local standard of rest (lsr) reference frame (Kerr & Lynden-Bell 1986), to range from −5 to 20 km s⁻¹ for both single stars and binaries to limit the contamination from the sources that are not the members of these clusters (T15, also more in Section 5), then MF becomes 5.8 ± 1.1% for ONC (30 multiples out of 518 stars observed) and 5.3 ± 1.2% for NGC 2264 (21 out of 397). Uncertainties were obtained as ${N}_{\mathrm{multiple}}^{\frac{1}{2}}{N}_{\mathrm{total}}^{-1}$.

To compare the MF that we observe toward ONC and NGC 2264 to that observed in the field, we ran a Monte-Carlo simulation producing a synthetic field population that would be consistent with the distribution of binary properties in the nearby G dwarfs (Raghavan et al. 2010). For each of the stars in our sample that had $-5\lt {v}_{\mathrm{lsr}}\lt 20$ km s⁻¹ and >3 detections (i.e., those from which the MF was measured), we generated randomly configured systems (including both single and multiple) that had the same dates of observation and the same uncertainties as the data. We then ran the same detection test that would identify binary systems from the generated population, producing an MF for a single test case. This process was repeated 1000 times. The average MF from all of the test cases, both in the ONC and NGC 2264 regions, was determined separately. Uncertainty was determined from the 1σ dispersion in the generated MF between the 1000 test cases.

For the ONC, there are 518 systems that make up the population from which the MF was estimated. Of these systems, ∼120 stars have uncertainties too large to be detected as variable in any of the runs. The average MF of the synthetic population is 4.8 ± 0.9%. NGC 2264 had a total of 397 systems, of which ∼70 cannot be identified as multiples in any run, and an average MF of 6.1 ± 1.2%.

The estimate of the expected MF makes an implicit assumption concerning the mass of the primary stars. To simulate the population of the field stars to be consistent with what was measured by Raghavan et al. (2010), 1 M_⊙ primaries were chosen. However, the masses of the stars monitored in these clusters are expected to be significantly lower. The typical effective temperature template matched toward these stars is 4000 K (Figure 2). From this, we can estimate them to be K stars with typical masses on the order of 0.7 M_⊙ (Baraffe et al. 2015). Stellar multiplicity varies strongly with stellar mass, and lower-mass stars have been found to have lower MF than higher-mass stars (Fischer & Marcy 1992; Duchêne & Kraus 2013). G stars have typically been used for comparison because surveys of their orbital parameters have been by far the most comprehensive.

Zoom In Zoom Out Reset image size

In addition, because the mass of the primary is lowered, the distribution of the masses of the secondary is also cut off at smaller values. Because of this, the difference in the peaks of the RV fluctuations due to the presence of a companion would become smaller as well; therefore, fewer systems would be detected. By generating a population with identical orbital parameters around 0.7 M_⊙, we instead infer MFs of 4.5 ± 0.9% for ONC and 5.8 ± 1.2% for NGC 2264.

The dispersion in the MF generated with identical orbital parameters between separate runs (all consisting of 1000 test cases) observed with the cadence and uncertainties set by the same population is typically <0.1%. To estimate the systematic effects due to a potentially different distribution of periods and overall number of binary systems in these clusters compared to what was previously found in the field, we varied one parameter at a time and looked for a difference in MF. Varying the underlying binarity fraction by 2% (1σ values quoted by Raghavan et al. 2010) typically changes the extracted MF by 0.2%. No uncertainties on the orbital parameters were made apparent, although varying the peak of the period distribution by 0.1 log P (where P is measured in days) produces an MF that is different by 0.5% when the average period is decreased from 293 to 233 years, and by 0.3% when the period is increased to 369 years. Varying the standard deviation of the period distribution by 0.1 ${\sigma }_{\mathrm{log}P}$ changed the MF by 0.4%. Changing the mass ratio and eccentricity from uniformly distributed to those that are described by Duquennoy & Mayor (1991) decreases the MF by 0.5% and 0.4%, respectively.

To determine the completeness limits that we probe with these observations, we recorded orbital parameters from all of the generated binary systems and determined a fraction that would be detectable relative to all of the binaries that satisfy the specific orbital parameters (Figure 3). Unsurprisingly, the most easily detectable systems have short orbital periods, comparable masses between the primary and the secondary, and an edge-on orientation. It is possible to detect only 60% of all of the systems with separations <1 au because either the other orbital parameters make a detection difficult or the uncertainty in the measurements that were applied to the generated velocity curves were too large for a reliable detection. Beyond the separations of 15 au, there are almost no systems that could be detected as binaries based on their RV variability. Combining all of the possible separations and inclination angles, only 15% of multiple systems can be detected for stars with a mass ratio on the order of unity (in this case both stars have mass of 1 M_⊙); however, this rapidly decreases to ∼5% for those companions with only 0.2 M_⊙.

Zoom In Zoom Out Reset image size

Individual cross-correlations of all of the systems identified as binaries were visually examined to determine whether or not it is possible to see a second peak due to the presence of the second star. We required that any star flagged as a double-lined binary must exhibit multiple peaks or a skewed correlation function in at least two epochs to minimize spurious detections. There were a total of 15 of such systems, of which 10 were found toward NGC 2264 and 5 toward ONC. In epochs where it was possible, a Gaussian was fitted to both peaks to find the velocities of both components. We assigned uncertainties to these measurements of 2 km s⁻¹, limited by the resolution of the extracted cross-correlations function. All of these measurements are included in Table 3.

Out of 130 sources flagged as binaries in this paper, 6 single-lined binaries have $\geqslant 10$ RV measurements and 2 double-lined binaries have $\geqslant 9$ measurements. We attempted to fit the velocity curves of these sources as they contained a sufficient number of measurements and redundancy on the measurements to obtain a unique solution. Because sources in NGC 2264 have been monitored more frequently than those in ONC (Figure 1), all 8 of these sources are found toward NGC 2264.

To fit the orbits, the IDL package RVFIT (Iglesias-Marzoa et al. 2015) was used, which fits the following quantities: P (the orbital period), T_P (the time of periastron passage), e (the eccentricity), ω (the argument of the periastron), γ (the systemic velocity of the system), and K₁ (the amplitude of the RV fluctuation) and K₂ (the amplitude of the RV fluctuation for the second star if the system in question is a double-lined spectroscopic binary.) The derived quantities are the semimajor axis ${a}_{1}\mathrm{sin}i$ and the binary mass function f. M sin i contributions of the individual stars can be distinguished for double-lined systems.

RV 1768 showed the second peak in the correlation in 7 out of 11 epochs, and both peaks exhibited a similar strength. The remaining 4 epochs had measurements very close to the mean of the remaining measurements. Because we are unable to distinguish between the contributions of individual components, there could be a larger spread in the velocity due to line blending; to those measurements we assign an uncertainty of 5 km s⁻¹.

As RV 1768 appeared to have almost equal mass in both components, there was some difficulty in distinguishing which of the two lines in a given epoch belonged to which star in cases of resolved measurements, and the velocity of which star was the most dominant in cases of unresolved measurements. An attemp was made for all of the perturbations of the line combinations to be fitted and examined by eye—the combination which provided the best fit is listed in Table 3 and the resulting fit is shown in Figure 4 and Table 5. The masses of the individual components (M  sin³i) are ∼0.25 M_⊙. Given that the spectra for this system are best fit by a ∼4000 K template, this is a reasonable fit requiring only a modest inclination angle.

Zoom In Zoom Out Reset image size

RV 1659 had a double-peaked correlation function in 8 out of 9 epochs, with one component clearly dominating over the other. It was best fit by a circular orbit, a period of 15.3 days, and M sin³i of 0.73 and 0.58 M_⊙ (Figure 5, Table 5). For orbits with no eccentricity, the argument of periastron carries no meaningful information.

Zoom In Zoom Out Reset image size

Fitting information on the six single-lined systems is presented in Table 6 (Figures 6–11). Based on the characteristic heliocentric velocity of RV 1372 of −32 km s⁻¹ (v_lsr ∼ −47 km s⁻¹), it is unclear whether or not it is a member of NGC 2264. However, this system is typically best fit by a 4250 K template and has near-infrared fluxes of J = 12.270, H = 11.718, and K = 11.464 (Cutri et al. 2003). While there is some contamination in the fluxes from the companion, they are in a good agreement with those of other binary systems which follow the same templates (e.g., RV 1550, a system for which the orbital fit is also available and which has a systematic velocity that is consistent with the cluster mean, has J = 12.262, H = 11.566, K = 11.423), making it likely that RV 1372 is not a foreground or background system, but rather that it was ejected from the cluster.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

RV 1753 was previously monitored for change in RV by Karnath et al. (2013) over a period of 20 years with 35 epochs. All of the fitted orbital parameters from that study are in excellent agreement with the fits presented in this paper.

Some of the analysis of the ONC region performed in T09 was affected by contamination from moonlight, and the lowest signal-to-noise data also added scatter. While the overall conclusions of T09 remain unchanged, the sample presented in this paper has higher quality velocity measurements. For this reason, once again, we look at the relation between stellar RVs and ¹³CO gas (Figures 12 and 13). All of the velocities in the plots are in the kinematic LSR reference frame to match that of the gas (Bally et al. 1987). To confirm that no binary stars contaminate the sample, we required that there be no variability in velocities with at least 3 epochs of observations. Unlike in T09, the peak of the RV distribution for stars follows the gas with no offset and a comparable velocity dispersion of ∼2.5 km s⁻¹. The only exception to this is the $-5\_\_AMP\_\_fdg;5\lt \delta \lt -5\_\_AMP\_\_fdg;0$ range, which coincides with the location of inner ONC regions such as Trapezium and OMC 2/3. A recent paper by Da Rio et al. (2015) measuring the RV from the infrared spectra in Orion A also found a lack of a blueshifted tail extending beyond v_lsr < 0 km s⁻¹.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

However, while not quite as pronounced as  reported by T09, we do observe some blueshifted tail component in the stellar RVs relative to the ¹³CO motions. To determine whether or not the sources that populate that tail are members of the cluster or unrelated foreground or background stars, we searched for Li i detection toward some of them (Table 4, Figure 12) as a signature of their youth to establish whether or not these sources could be causally related to the ONC. Overall, many of the sources that occupy the same velocity space as the gas and blueshifted tail were indeed found to contain Li i. While there is some contamination from sources that appear to be somewhat more evolved, their low numbers alone cannot account for the entirety of the blueshifted tail. On the other hand, nearly all of the sources that occupy the velocity space outside what is presented in Figure 12 lack in Li i.

Table 4.  Sources Surveyed for the Presence of Li i

RV	R.A.	Decl.	vLib	${\sigma }_{\mathrm{Li}}$	RLi	${\lambda }_{\mathrm{Li}}$ c	WLid	Temp
#a	(J2000)	(J2000)	(km s−1)	(km s−1)		(Å)	(Å)	(K)
113	05:34:15.45	−06:06:55.1	49.51	0.72	10.58	...	...	3750
120	05:34:17.78	−05:55:43.1	4.52	0.99	7.69	...	...	6000
133	05:34:20.80	−05:23:29.2	18.62	1.48	4.87	6708.32	0.39	3500
160	05:34:27.34	−05:24:22.2	26.02	2.26	9.18	6708.46	0.47	4000
164	05:34:28.22	−05:59:09.0	33.56	0.82	10.78	6708.62	0.10	6250

Notes.

^aSources with RV # greater than 2057 have $R\lt 6$ for all detections, and thus they are not included in the Tables 2 and 3. ^bVelocity and other properties were measured only from Li i data. ^cTypical uncertainty in ${\lambda }_{\mathrm{Li}}$ is 0.01 Å. ^dTypical uncertainty in W_Li is 0.01 Å.

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

Table 5.  Orbital Parameters for Double-lined Binaries

Parameter	RV 1768	RV 1659
Adjusted Quantities
P (day)	4.7878 ± 0.0002	15.3182 ± 0.0007
Tp (HJD)	2454289.50 ± 0.06	2454294.07 ± 0.05
e	0.55 ± 0.05	0.00 ± 0.02
ω (deg)	139.68 ± 3.16	0.00 ± 0.60
γ (km s−1)	16.05 ± 0.54	22.83 ± 0.46
K1 (km s−1)	59.34 ± 6.34	41.56 ± 1.07
K2 (km s−1)	60.83 ± 6.32	52.20 ± 1.05
Derived Quantities
${M}_{1}{\mathrm{sin}}^{3}i$ (M⊙)	0.254 ± 0.066	0.73 ± 0.04
${M}_{2}{\mathrm{sin}}^{3}i$ (M⊙)	0.248 ± 0.065	0.58 ± 0.03
$q={M}_{2}/{M}_{1}$	0.98 ± 0.15	0.80 ± 0.03
${a}_{1}\mathrm{sin}i$ (106 km)	3.26 ± 0.37	8.75 ± 0.23
${a}_{2}\mathrm{sin}i$ (106 km)	3.35 ± 0.37	11.00 ± 0.22
$a\mathrm{sin}i$ (106 km)	6.61 ± 0.52	19.75 ± 0.32
Other Quantities
${\chi }^{2}$	8.38	13.43
Nobs (primary)	10	9
Nobs (secondary)	8	8
Time span (days)	2308.7	2308.6
rms1 (km s−1)	1.62	1.22
${\mathrm{rms}}_{2}$ (km s−1)	1.42	2.10
Tave (K)	4021	4203

Download table as:  ASCIITypeset image

Table 6.  Orbital Parameters for Single-lined Binaries

Parameter	RV 1166	RV 1372	RV 1496	RV 1550	RV 1660	RV 1753
Adjusted Quantities
P (day)	105.82 ± 0.27	315.41 ± 5.79	588.93 ± 9.47	72.83 ± 0.03	622.00 ± 26.32	12.93 ± 0.01
Tp (HJD)	2452620.5 ± 29.3	2454382.4 ± 5.5	2454814.5 ± 22.1	2454329.0 ± 1.1	2454585.1 ± 31.4	2454299.3 ± 0.6
e	0.00 ± 0.07	0.65 ± 0.12	0.44 ± 0.07	0.32 ± 0.02	0.56 ± 0.08	0.170 ± 0.03
ω (deg)	224.02 ± 99.11	178.80 ± 5.31	8.61 ± 11.29	266.43 ± 7.12	135.59 ± 11.99	194.66 ± 19.74
γ (km s−1)	15.93 ± 0.25	−29.97 ± 0.78	28.05 ± 0.30	20.80 ± 0.26	23.08 ± 0.39	17.97 ± 1.35
K1 (km s−1)	4.24 ± 0.46	20.00 ± 7.73	3.46 ± 0.87	16.91 ± 0.39	5.86 ± 0.51	16.11 ± 1.61
Derived Quantities
${a}_{1}\mathrm{sin}i$ (106 km)	6.17 ± 0.67	65.91 ± 27.09	25.09 ± 6.38	16.06 ± 0.39	41.36 ± 4.93	2.82 ± 0.28
$f({m}_{1},{m}_{2})$ (M⊙)	0.0008 ± 0.0003	0.11 ± 0.15	0.0018 ± 0.0014	0.031 ± 0.003	0.007 ± 0.003	0.005 ± 0.002
Other Quantities
χ2	1.31	2.61	1.74	1.76	2.61	2.61
Nobs (primary)	10	11	10	11	11	11
Time span (days)	903.6	2308.7	2327.7	2325.7	2325.6	2327.7
rms1 (km s−1)	0.17	0.53	0.19	0.25	0.55	0.28
Tave (K)	6274	4267	5577	4361	4256	5301

Download table as:  ASCIITypeset image

There are several possible explanations for this tail. There could be a separate foreground population of young stars that is not an immediate part of the ONC. Alves & Bouy (2012) and Bouy et al. (2014) argue that NGC 1980 is an example of such a foreground cluster. Unfortunately, it is located at 837 < α < 839 and −61 < δ < −58, and the presence of the blueshifted population in that region is minimal and not spatially coherent. This is consistent with what has been found by Da Rio et al. (2015). However, south of it, the ONC is starting to turn into the L1641 cloud, and thus it is possible that the tail in the southernmost regions can be attributed to this.

Some of these stars could have been dynamically scattered to achieve these velocities. While it is difficult to explain why there is no redshifted population to make the velocity distribution symmetric, it is possible that high extinction prevents us from observing it. The inner ONC is where this effect would be the most pronounced. Not only does this region have significantly higher stellar density than the rest of the cluster, allowing for more significant dynamical interactions between stars, but it is also more affected by extinction due to the high density of gas. It is possible that this can account for some of the observed blueshifted sources.

Alternatively, it is possible that the gas was being blown away by stellar feedback, leaving a somewhat older population of stars behind while newer stars formed. As suggested by T09, it is likely that in the northmost region, in the vicinity of NGC 1977, gas has been pushed back by irradiation from HD 37018, HD 37077, and HD 36958, which are B1V, B3V, and B3V stars, respectively, leaving behind a mini-cluster.

Proszkow et al. (2009) and T09 instead postulate that the redshift in the gas in Trapezium and OMC 2/3 could be due to the gravitational infall of the OMC 2/3 filament toward the Trapezium cluster. However, this would not entirely explain the presence of the blueshifted stellar population relative to the gas, and the location of these blueshifted stars is not correlated with either being on or off of the filament. More precise distances and proper motions, which could in part be obtained by the ongoing Gaia mission, are needed to confirm or deny the infall of the OMC 2/3.

While there is little substantial difference between the velocities for the NGC 2264 region quoted in T15 and this paper, some improvements could be made to the previous analysis of the velocity structure for the region through better filtering of the spectroscopic binaries. Similar to the case of the ONC, we restrict our analysis only to those sources that were detected in at least three epochs and show no RV variability. The position–velocity diagram for the stars is compared to that for gas from Ridge et al. (2003) in Figures 14 and 15.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We impose a constraint on sources to have an R.A. of $100.05\geqslant \alpha \geqslant 100.4$ in order to only trace those objects that are spatially correlated with ¹³CO gas to limit contamination from foreground or background sources. As a result, stars that are located north of δ ∼ 955 (i.e., Spokes Cluster and S Mon) have an agreement in RV with that of the gas that is significantly better than what was presented before by T15. This is partially due to the stricter spatial constraints than were originally imposed. No objects exhibit a significant blueshift in RV relative to the gas, with the slight exception of the southernmost declinations in the Spokes Cluster.

However, the entirety of the stellar population found toward the Cone Nebula does show a significant blueshift that is not dissimilar to what is found toward the Trapezium and OMC 2/3 regions in the ONC. However, unlike Trapezium where the dispersion velocity of the stars is wide enough to also correlate with the gas, the stars toward the Cone Nebula appear to be decoupled from the gas. The reason for this is not entirely clear.

In addition to these regions, there appears to be a small cluster of stars centered at α ∼ 10045, δ ∼ 97 with a diameter of ∼01 (Figure 14). It was previously identified but not discussed in T15. The members of this cluster appear to have a systematic RV_lsr ∼ 2 km s⁻¹, which is somewhat distinguishable from the main cluster. It is possible that it is a an older cluster that has managed to clear away all of its gas.

Since this cluster does not appear to be dynamically relaxed, as it exhibits a significant distinct substructure, a better determination of its age and further modeling will be needed to determine the degree of interaction between these subclusters. This could shed light on the dominant method of cluster formation, such as whether or not it is undergoing cold collapse (Scally & Clarke 2002).