Abstract
Long-term monitoring observations with the Hitachi 32 m radio telescope of the 6.7 GHz methanol masers associated with the high-mass star-forming region G5.900−0.430 are presented. A period of flux variability at approximately 1260 d is detected in the features at VLSR = 9.77 and 10.84 km s−1, while a secondary shorter period, 130.6 d, is determined for the 0.66 km s−1 feature. This is only the second source which has two different periods. The period of ∼1260 d is approximately twice as long as the longest known period of 6.7 GHz methanol masers. The variability pattern of the symmetric sine curves and the consistency with the expected period–luminosity relation suggest that the mechanism of the maser flux variability of the 9.77 and 10.84 km s−1 features in this source can be explained by protostellar pulsation instability. On the other hand, because the 0.66 km s−1 feature has an intermittent and asymmetric variability profile, we propose that this feature is explained by the colliding wind binary or spiral shock models. Obtaining the spatial distribution of the 0.66 km s−1 feature using very long baseline interferometry will lead to a better understanding of this source.
1 Introduction
High-mass stars have significant impact on their surrounding environment through various feedback mechanisms, e.g., stellar winds, ultraviolet radiation, and supernovae. However, our understanding of the formation processes of high-mass stars remains inadequate, hampered by the observational difficulties of their birth in distant, very deeply embedded dense gas. The Class II 6.7 GHz methanol maser is a well-known tracer of high mass star-forming regions (HMSFRs) (e.g, Menten 1991; Minier et al. 2003; Pandian et al. 2007; Breen et al. 2013). Emission arises in most cases from regions of an interface between the protostellar disk and envelope at a typical distance of ∼1000 au from a protostar (Bartkiewicz et al. 2016). The 6.7 GHz maser is sensitive to the local physical conditions around high-mass protostars (e.g., Cragg et al. 2005; Sugiyama et al. 2008; Moscadelli et al. 2017; Burns et al. 2020), and thus is an excellent observational probe to study the high-mass star formation process.
Goedhart, Gaylard, and van der Walt (2003) first discovered a periodic flux variation of 243.3 d in the 6.7 GHz methanol masers associated with HMSFR G9.62+0.20E. Another 27 periodic methanol maser sources have been reported (Goedhart et al. 2004, 2009, 2014; Szymczak et al. 2011, 2014, 2015, 2016; Araya et al. 2010; Fujisawa et al. 2014; Maswanganye et al. 2015, 2016; Sugiyama et al. 2015, 2017; Proven-Adzri et al. 2019; Olech et al. 2019, 2022), and their periods range from 23.9 to 668 d. Systematic monitoring observations towards large samples have contributed significantly to the discovery of maser periodicity (e.g., Goedhard et al. 2004; Sugiyama et al. 2018; Szymczak et al. 2018; Olech et al. 2022). The variation pattern in the time series of the 6.7 GHz methanol maser are classified into two types: continuous (e.g., G188.95+0.89 reported by Goedhard et al. 2014) and intermittent (e.g., IRAS 22198+6336 = G107.298+0.5639 reported by Fujisawa et al. 2014). Such differences in variation patterns are thought to be due to differences in the origin of the 6.7 GHz maser periodicity. The mechanism of the maser periodicity remains an open question, and several explanations have been proposed; a colliding wind binary (CWB) system (van der Walt 2011; van der Walt et al. 2016), protostellar pulsation (Inayoshi et al. 2013), spiral shock (Parfenov & Sobolev 2014), periodic accretion in a circumbinary system (Araya et al. 2010), or a very young low-mass companion blocking the ultraviolet radiation from the high-mass star in an eclipsing binary (Maswanganye et al. 2015). These models can explain the maser fluctuations of some sources well; however, there is no clear consensus.
In this paper, we present the new discovery of periodicity in the 6.7 GHz methanol maser source in HMSFR G5.900−0.430, and discuss the mechanism of its flux variability.
2 Observation
Monitoring observations of the 6.7 GHz methanol maser were made with the Hitachi 32 m telescope of Ibaraki station, a branch of the Mizusawa VLBI Observatory of the National Astronomical Observatories Japan (NAOJ), operated jointly by Ibaraki University and NAOJ (Yonekura et al. 2016). This is as a part of the Ibaraki 6.7 GHz Methanol Maser Monitor (iMet) program.1
Monitoring observations of G5.900−0.430 began on 2013 January 03 [modified Julian day (MJD) = 56295]. The cadence of observations is once per every ∼10 d from the start of the monitoring observations until 2015 August 08 (MJD = 57242), and once per every ∼5 d from 2015 September 19 (MJD = 57284) to the present. Observations after 2014 May 08 (MJD = 56785) were made at about the same azimuth (∼187°) and elevation angle (∼27°) to minimise intensity variations due to systematic telescope pointing errors. The half-power beam width of the telescope is |$\sim {4{_{.}^{\prime}}6}$| with a pointing accuracy better than |$\sim {30^{\prime\prime}}$|. The coordinates of G5.900−0.430 adopted in observations are |$\mbox{RA} = {18^h00^m 40{_{.}^{\rm s}}86}$|, |$\mbox{Dec} = {-24^{\circ}04^{\prime }20{_{.}^{\prime\prime}}8}$| (J2000.0) (Caswell 2009; Caswell et al. 2010).
Observations are made by using a position-switching method. The OFF position is set to ΔRA = +60′ from the target source. The integration time per observation is 5 min for both the ON and OFF positions. A single left-hand circular polarization (LCP) signal was sampled at 64 Mbps (16 mega-samples per second with 4 bit sampling) by using a K5/VSSP32 sampler. The recorded bandwidth is 8 MHz (RF: 6664–6672 MHz) and they are divided into 8192 channels. After averaging over three channels, the 1σ rms noise level is approximately 0.3 Jy and the velocity resolution is 0.13 km s−1. The antenna temperature was measured by the chopper-wheel method and the system noise temperature toward the zenith after the correction for the atmosphere opacity (|${T}^{*}_{\rm sys}$|) is typically 25–35 K. In our monitoring program, we observe ∼ 60 methanol maser sources per day, of which the variation of the flux density of sources that do not show the intrinsic variation is less than |$\sim \!\! 20\%$|.
3 Result
The averaged spectrum and dynamic spectrum compiling all 437 scans of G5.900−0.430 is presented in figures 1 and 2 . In this source, seven velocity features are distributed from VLSR ∼0 km s−1 to ∼14 km s−1. According to Caswell et al. (2010), the 6.7 km s−1 feature (C and D in figure 1) is associated with HMSFR G5.885−0.393, approximately |${2{_{.}^{\prime\prime}}3}$| apart from G5.900−0.430. Therefore we exclude these features from the discussion below.
Averaged spectrum of the 6.7 GHz methanol maser associated with G5.900−0.430. All 437 scans of 5 min duration are averaged and the 3σ detection limit is 0.042 Jy. Labels A, B, C, D, E, F, and G indicate spectral components at VLSR = 0.66, 4.35, 6.76, 7.29, 9.77, 10.84, and 14.02 km s−1 , respectively.
Dynamic spectrum compiling all 437 scans of the 6.7 GHz methanol masers associated with G5.900−0.430. Labels on the upper axis indicate the beginning of each year. White blanks indicate no observations were made.
Time series and results of LS analysis for all velocity components in G5.900−0.340. The left-hand column shows the time series of flux density of each velocity feature. We excluded the data points with flux densities of less than 3σ. The second and third columns show the Lomb–Scargle power spectra plots. The dotted lines in the second and third columns represent the |$0.01\%$| false alarm probability (probability of judging noise as a real signal) levels; if the peak value of the power spectrum is higher than the dotted line, the obtained period is reliable.
Results of power function fitting. Gray points represent time series of each periodic component excluding the data points with flux densities less than 3σ, and black lines show the best fitting of periodic power function. It should be noted that the entire data were used for the fitting, but for the 0.66 km s−1 feature (left-hand panel) we show here only the data from MJD = 59290 to MJD = 59530.
Parameters of the periodicity.
| VLSR | PLS* | Pfit† | f‡ | r2§ |
|---|---|---|---|---|
| (km s−1) | (d) | (d) | ||
| 0.66 | 130.6 (2.8) | 130.5 (0.1) | 0.73 (0.07) | 0.44 |
| 9.77 | 1264.9 (265.5) | 1260.3 (8.6) | −0.20 (0.07) | 0.67 |
| 10.84 | 1264.9 (324.3) | 1260.7 (9.6) | −0.11 (0.07) | 0.68 |
| VLSR | PLS* | Pfit† | f‡ | r2§ |
|---|---|---|---|---|
| (km s−1) | (d) | (d) | ||
| 0.66 | 130.6 (2.8) | 130.5 (0.1) | 0.73 (0.07) | 0.44 |
| 9.77 | 1264.9 (265.5) | 1260.3 (8.6) | −0.20 (0.07) | 0.67 |
| 10.84 | 1264.9 (324.3) | 1260.7 (9.6) | −0.11 (0.07) | 0.68 |
Period estimated by LS method.
Period estimated by power function fitting.
Asymmetric parameter.
Coefficient of determination.
Parameters of the periodicity.
| VLSR | PLS* | Pfit† | f‡ | r2§ |
|---|---|---|---|---|
| (km s−1) | (d) | (d) | ||
| 0.66 | 130.6 (2.8) | 130.5 (0.1) | 0.73 (0.07) | 0.44 |
| 9.77 | 1264.9 (265.5) | 1260.3 (8.6) | −0.20 (0.07) | 0.67 |
| 10.84 | 1264.9 (324.3) | 1260.7 (9.6) | −0.11 (0.07) | 0.68 |
| VLSR | PLS* | Pfit† | f‡ | r2§ |
|---|---|---|---|---|
| (km s−1) | (d) | (d) | ||
| 0.66 | 130.6 (2.8) | 130.5 (0.1) | 0.73 (0.07) | 0.44 |
| 9.77 | 1264.9 (265.5) | 1260.3 (8.6) | −0.20 (0.07) | 0.67 |
| 10.84 | 1264.9 (324.3) | 1260.7 (9.6) | −0.11 (0.07) | 0.68 |
Period estimated by LS method.
Period estimated by power function fitting.
Asymmetric parameter.
Coefficient of determination.
A periodic and intermittent profile is found at 0.66 km s−1 feature (A in figure 1). Periods of 130.6 ± 2.8 and 65.2 ± 0.7 d were determined from the LS periodicity analysis. Note the latter appears to be a harmonic of the former, i.e., P = 130.6/n where n = 2. In fact, in figure 5, almost no flares can be detected in the period of 65.3 d (dashed line), which is half of 130.6 d (solid line). Thus 130.6 d is the most plausible period for the 0.66 km s−1 feature. An asymmetric power function fitting determines the value of Pfit = 130.5 ± 0.1 d and f = 0.73. The periods obtained by the two methods are within the errors measured. The value of f suggests that the 0.66 km s−1 feature experiences an asymmetrical temporal flux density variation; a rapid onset followed by a slower decline. The HWHM of rising and decaying time obtained directly from fitting results are 8.16 and 13.88 d, respectively.
Time series of 0.66 km s−1 feature. The solid lines show every 130.6 d from MJD = 56342.3, which is the first peak of the best-fitting power function and the dotted lines are in the middle of the solid lines. Labels on the upper axis indicate the beginning of each year. The thick solid lines around MJD = 56734, 56865, and 59607 indicate that a flux density over 3σ was not detected during the eight days before and after the expected flare date.
As shown in figure 5, during our monitoring observation, there were three cases where no flares were detected at the expected flare date. In the first two of these three cases, observations were not made within five days before and after the expected flare date due to maintenance or other reasons. In the third case, we detected emissions with flux density of 2.5σ and 2.6σ at two days before and three days after the expected flare date (MJD = 59607), respectively, but no emissions above 3σ.
The 4.35 km s−1 feature (B in figure 1) is generally below detection limits after MJD = 57363, and no periodicity is detected for this feature.
The 9.77 km s−1 and 10.84 km s−1 features (E and F in figure 1) experience very long and continuous flux variation. The estimated period from the LS method and power function fitting for the 9.77 km s−1 feature are 1264.9 ± 265.5 d and 1260.3 ± 8.6 d, respectively, and those for the 10.84 km s−1 feature are 1264.9 ± 324.3 d and 1260.7 ± 9.6 d, respectively. The periods of each velocity feature obtained by the two methods are consistent within the estimated errors. The results of the fitting show that the flux variation of these features are symmetric with f = −0.20 and −0.11 for 9.77 and 10.84 km s−1, respectively.
MacLeod et al. (2022) detected periodic flux variation of maser features with two different periods in HMSFR G9.62+0.20E. According to MacLeod et al. (2022), G9.62+0.20E has secondary period of 52 d for 8.8 km s−1 feature, along with the previously reported period of 243 d for another features. G5.900−0.430 may be the second example which has a secondary period. The temporal profiles of the 8.8 km s−1 feature and the others are quite different; intermittent and continuous, respectively. This is similar to the present results for G5.900−0.430. On the other hand, the relationship in the flux variation profiles and the periods shows a completely opposite trend. In G5.900−0.430, the feature with a long period shows a continuous profile and the feature with a short period shows an intermittent profile, whereas G9.62+0.20E has the opposite tendency.
In the 14.02 km s−1 feature (G in figure 1), only 24 of the 437 scans have a flux density over 3σ and no periodicity was detected. Because of the small number of data points, this feature is not addressed in the following discussion.
4 Discussion
4.1 Spatial distribution of maser features
The 6.7 GHz methanol maser emission of G5.900−0.430 was imaged by Hu et al. (2016) with the Karl G. Jansky Very Large Array (VLA). The 0.66 km s−1 feature was not detected and 4 km s−1 to 10 km s−1 features were imaged. The observations by Hu et al. (2016) were made between MJD = 55985 and 56033. Assuming that the 0.66 km s−1 feature has a period of 130 d before we started the monitoring, the expected flare dates closest to dates of observations by Hu et al. (2016) are MJD = 55952 and 56082, both of which are outside the observations of Hu et al. (2016). Thus the observations of Hu et al. (2016) must have been conducted during the quiet state of the 0.66 km s−1 feature. According to Hu et al. (2016), the 9–11 km s−1 features have a special distribution of ∼600 mas, which correspond to ∼1700 au when the distance of 2.9 kpc is adopted (see subsection 4.2). This is approximately comparable in size to typical maser distribution in high-mass protostar system. Therefore, the high-velocity features with a long period are thought to be excited by the central protostar.
The 4.35 km s−1 feature is only just above our detection limits in our most recent single-dish observation. These masers are superimposed on a bright, 7.4 Jy, H ii region, obtained by two 1024 MHz sub-bands from 4.9840 to 6.0080 GHz and from 6.6245 to 7.6485 GHz, suggesting they are associated (Hu et al. 2016).
4.2 Periodicity
According to Olech et al. (2019, 2022), among the known periodic maser sources, the longest period is 668 d in G196.45−1.68 (Goedhart et al. 2004) and only two sources have periods longer than 500 d. Our result increases the upper end of the range of known periods by a factor of two. On the other hand, the 0.66 km s−1 feature has a much shorter period, 130.6 d, and the flaring is intermittent, perhaps owing to the relative weakness of the feature. The difference in the behavior of the variability of the low-velocity feature (0.66 km s−1) and high-velocity features (9.77 and 10.84 km s−1) suggests that the variability of the flux of these components may be the result of different mechanisms. We discuss the mechanisms of maser variability below.
The CWB model which causes periodic variations in HMSFRs was first proposed by var der Walt, Goedhart, and Gaylard (2009), and it was modeled more in detail by van der Walt (2011). In this model, periodic wind interaction in a binary system generates changes in the background free–free emission which is amplified by the masers. It requires a varying H ii region, where the free–free radiation acts as the seed photons. According to van der Walt et al. (2011, 2016) and Olech et al. (2022), the maser flare temporal profile may be explained by the CWB model where it is typified by a short onset and long decay. This is the best explanation for the flare characteristics of the 0.66 km s−1 feature in G5.900−0.430. It is proposed that resulting variations of the associated background H ii region cause the flaring in the 0.66 km s−1 feature. However, this model is unsuitable for the high-velocity features, which show continuous sinusoidal variation.
A rotating spiral shock model in a binary system is proposed by Parfenov and Sobolev (2014). The dust temperature variations are caused by the rotation of hot and dense material in the spiral shock wave in the circumbinary disk central gap. This model does not require either radio or infrared emission variability, but needs an edge-on protostellar disk with which masers are associated. According to van der Walt et al. (2016), the flux variation pattern in the spiral shock model has to show a tail in the decaying phase, resulting in a non-quiescent phase and difficulty in presenting an intermittent profile of each flare. Morgan et al. (2021) suggests that the source geometry and orientation are also important factors influencing observed flare profiles. Therefore, it is important to clarify the spatial distribution of maser features and their flux variation pattern.
Unfortunately, the spatial distribution of this source, especially for VLSR < 4 km s−1, is not observationally revealed and we cannot examine the validity of this model. Therefore, we cannot judge whether the spiral shock model is appropriate for the flux variation of the 0.66 km s−1 feature or not. High-resolution very long baseline interferometry (VLBI) observations during the active phase of the 0.66 km s−1 feature are required to understand the periodic variation of this source better.
In general, however, pulsation of the central star spreads spherically without directionality; therefore, in the pulsation model all features will vary with the same period. On the other hand, for the CWB and spiral shock models, the presence or absence of periodic flux variation, and the phase of periodic flux variation, may vary from feature to feature, depending on their geometric arrangement. Thus the 0.66 km s−1 and high-velocity features may be associated with different sources. VLBI observations are required to obtain the spatial distribution of the 0.66 km s−1 and high-velocity features to resolve this open question.
5 Summary
We present the new periodic 6.7 GHz methanol maser source G5.900−0.430, detected from long-term monitoring observations conducted by Ibaraki University. Periodic flux variabilities was detected in three velocity features out of five in this source. The obtained periods are 130.6 d for one low-velocity feature and ∼1260 d for two high-velocity features. The period of 1260 d is the longest ever found. From the intermittent and asymmetric variability profile, the low-velocity feature is likely to be explained by the CWB model or spiral shock model, while the continuous and symmetric variation of the high-velocity features are likely to be caused by the protostellar pulsation. Consistency with the expected period–luminosity relation also supports the pulsation as the cause of the periodic variation of the high-velocity components. Simultaneous VLBI observations of the high-velocity and low-velocity components during the active phase of the 0.66 km s−1 feature will lead to a better understanding of the nature of this source.
Acknowledgements
The authors are grateful to all the staff and students at Ibaraki University who have supported observations of the Ibaraki 6.7 GHz Methanol Maser Monitor (iMet) program. This work is partially supported by the Inter-university collaborative project “Japanese VLBI Network (JVN)” of NAOJ and JSPS KAKENHI Grant Numbers JP24340034, JP21H01120, and JP21H00032 (YY).
