5. Description of the individual objects

1. Supergiants (SGs)

1. VY CMa (Fig. 2)

The spectral type of this star is M5 and its optical variability can reach an amplitude of 3 mag. In general this object is the strongest SiO emitter among our targets. The observations of VY CMa by Lane (1982) during 1979-1980 yielded however weaker v=1 and v=2, J=1-0 lines than in VX Sgr, W Hya, R Leo, and Orion IRc2. During our monitoring, the observed SiO maser emission of this object was always intense in the two transitions. The line profile was very stable showing several spikes, the strongest one centered at 22.5 $\pm$ 0.5 km$\,$s$^{\rm -1}$ for the two transitions. The velocity of this spike agrees with the one of two vibrationally excited H₂O masers ( $\nu_{2}=1,~4_{4,0}$-5_3,3 and $\nu_{2}=1,~5_{5,0}$-6_4,3) observed by Menten & Melnick (1989). These two water vapor masers have high excitation energy (3065 and 3462 K respectively) similarly to our SiO masers (1750 and 3500 K respectively for the v=1 and v=2 lines). The dominant spike in the SiO spectra published by Lane (1982) showed a different velocity, $\sim20-21$ km$\,$s$^{\rm -1}$.

In our monitoring, the SiO variations of this object show a low contrast, i.e. a low peak flux (or total integrated area) ratio between consecutive maximum and minimum epochs, estimated to be $\sim1.3$. The two SiO maxima found in our observations occur more or less in phase with two optical maxima. For the peak flux curve of the v=1 line, we could find a regular variability with a period of 1000-1200 days but the minimum at JD $\sim$ 2447700 is not clear in the corresponding integrated flux curve. For v=2, the flux curves do not clarify the situation. Despite of this, the optical data seem to show a period of $\sim$ 1300 $\pm$ 200 days in good agreement with the tentative period for the SiO variability. On the contrary, Herman & Habing (1985) derived a period of 989 days from observations of the 1612 MHz OH maser emission, but note that this monitoring was not simultaneous with ours. The v=2/v=1 peak flux and integrated flux ratios are low: 0.65 and 0.55 respectively. This result is similar for other SGs. The centroid of the two lines has shown a secular shift of +2 km$\,$s$^{\rm -1}$ during the 2000 days of the monitoring, this variation being mainly due to intensity changes in the secondary spikes. On average, the equivalent widths (the integrated area to peak flux ratios) of the v=1 and v=2 lines are respectively 5.5 and 4.8 km$\,$s$^{\rm -1}$; as it is usual in SGs (Alcolea et al. 1990) these figures are significantly higher than for red giants (Miras or semiregulars).

2. $\mu$ Cep (Fig. 3)

This star is an M2 supergiant with a small optical variation amplitude ( $\Delta m_{v}=1.7$), and a possible period of about 730 days. Our monitoring for this star covers $\sim$ 1000 days. We see a similar increase in both v=1 and v=2 peak flux cuves by a factor of 5 to 6. During the last part of the monitoring (200-300 days) the intensity has been almost constant. If this is the beginning of a decrease in the intensity and there is a regular behavior, the period of variability should be at least of 1500 days (in disagreement with the proposed optical period). The integrated flux curves show a similar behavior. In general the value of the v=2/v=1 integrated flux ratio is low, $\sim$ 0.8, except for the epoch of minimum emission (JD  $2\,446\,959-2\,447\,050$) where it reaches a value of 1.3.

The line profiles of the two transitions look always very similar (one spike over a wider component). The velocity of the spike has increased 3 km$\,$s$^{\rm -1}$ in 1000 days, although the shift of the velocity centroid has been less than that. The v=1 equivalent width was $\sim3.5-4$ km$\,$s$^{\rm -1}$ at the beginning of the monitoring. It went down to 2.5 km$\,$s$^{\rm -1}$ at JD 2447100, just in phase with the curve of the v=2/v=1 integrated flux ratio. The v=2 equivalent width shows a similar behavior but less pronounced. In general, the values of the v=1 and v=2 equivalent widths for this star are lower than the typical values for supergiants.

3. VX Sgr (Fig. 4)

This object is an M4-M10 supergiant, which presents regular variations in the visible, with a period of 733 days and amplitudes $\Delta m_v$ $\sim$ 3 mag. The observed SiO variability is in general well correlated with the optical curve. The contrast is very low ($\sim$ 1.5) compared to those of Mira stars. The v=1 emission has a secular weakening of about a factor of 2 in 2000 days. The spectra obtained by Lane (1982) for this star show that this weakening process has been occurring since 1980 (a factor of 4-5 in total is deduced for 10 years). The short duration of the monitoring for v=2 does not allow to discuss those details for this line. The mean value of the peak flux v=2/v=1 ratio is $\sim$ 0.65; $\sim$ 0.45 for the integrated flux ratio. The mean values of the equivalent widths for the v=1 and v=2 lines are 10.5 and 8.3 km$\,$s$^{\rm -1}$, respectively. All these figures are typical of SGs. The line profiles are very complex and highly variable. In general they consist of 6 or 7 peaks at velocities between -5 and +20 km$\,$s$^{\rm -1}$. On average, the lifetime of the individual peaks is about 6 months, much less than the SiO period ($\sim$ 2 yr).

2. Semiregulars (SRGs)

1. GY Aql (Fig. 5)

The spectral type of this star is M6-M10. Its variability was not well known before the 1980's because GY Aql is weak in the optical. Now, its optical period has been established between 380 and 430 days. The amplitude of the optical variability is $\sim$ 6 mag (from $\sim$ 10 to 16 m_v). GY Aql is one of the strongest SiO emitters among SRGs. Its SiO variability is regular, with a period in agreement with the newly found optical period. The contrast is relatively high for a semiregular, $\sim~4-5$, a value typical of Miras. The v=2/v=1 integrated flux ratio is in general $\sim$ 1. The observed line profiles often show a single peak at 34 $\pm$0.5 km$\,$s$^{\rm -1}$, but the equivalent widths can change quite significantly (from 1 to 6 km$\,$s$^{\rm -1}$), with mean values of 3 and 2 km$\,$s$^{\rm -1}$ respectively for the v=1 and v=2 masers.

2. RT Vir (Fig. 6)

The optical variability of this object is practically unknown; very few data are available. The SiO maser emission is weak. The light curves for the integrated and peak fluxes and are noisy, but it is possible to see sharp changes in the intensities of the masers. In particular, note the strong increase in the SiO intensity 100-200 days after the optical burst occurred on JD 2447650. The mean value of the v=2/v=1 integrated flux ratio is 1.5. The line profiles show important changes in their structure. The velocity range where emission has been found is large: $\sim$ 18 km$\,$s$^{\rm -1}$. The v=1 and v=2 equivalent widths are, on average, 4.5 and 3.5 km$\,$s$^{\rm -1}$, respectively.

3. Mira variables

1. R Aqr (Fig. 7)

This object is a symbiotic system, consisting of a Mira variable and a hot dwarf companion. It is one of the rare examples among the members of this category where SiO maser emission has been detected. Probably, the molecular shell around R Aqr is very thin, since it has not been detected in CO emission (e.g. Young 1995). The interferometric observations of the v=1, J=2-1 SiO maser in this object by Hollis et al. (1990) showed that the maser position and that of the red giant are coincident within the error-bars of the measurements. So, one can expect that the SiO masers in this object are similar to those of other Mira variables, as it is confirmed by our data.

The SiO flux curves of this star show a good correlation with its optical variability. There is a phase lag between the SiO maximum and the optical maximum of 0.1-0.2periods, while SiO and IR vary in phase. The contrast is moderate for a Mira: $\sim3-4$. The intensity of the SiO masers shows a secular increase from JD 2446000 to JD 2448000. The v=2/v=1 integrated flux ratio has also been increasing during the same period from 0.8 to 2. The velocity centroid of the maser emission has shown variations (simultaneously in the two lines) up to 6-7 km$\,$s$^{\rm -1}$ that are not related with the optical phase. These velocity variations are associated to very important changes in the shape of the spectra (vanishing of some maser spikes and apparition of new spikes at very different velocities; see e.g. the v=1 spectra around JD 2446650 and 2447200). The most stable peak appears at $\sim$ 27.5 km$\,$s$^{\rm -1}$. The v=1 and v=2 equivalent widths show average values of 4.2 km$\,$s$^{\rm -1}$ and 3.8 km$\,$s$^{\rm -1}$, similarly to other Miras.

2. R Aql (Fig. 8)

Although the monitoring in object is very short, the data suggest that also in this Mira star the SiO variability follows the stellar cycle. The typical phase lag between the optical and SiO curves is not clear, however. The line profiles of the two transitions are in general similar, with a v=2/v=1 integrated flux ratio of less than 1.0 (equivalent widths $\sim$ 3.5 km$\,$s$^{\rm -1}$).

3. TX Cam (Fig. 9)

This giant star displays a very late spectral type, M8-M10. All SiO light curves show the characteristics of other Miras with the usual phase lag with respect to the visible. The behavior of the SiO light curves is quite regular, with a contrast of $\sim$ 6. From maximum to maximum we find differences up to a factor of 3 for the peak flux and up to a factor of 2 for the velocity integrated flux. The v=2/v=1 integrated flux ratio is $\sim$ 1 independently of the phase. The lines show the most important spike at $\sim$ 9 km$\,$s$^{\rm -1}$. On average, the v=1 equivalent width is 5 km$\,$s$^{\rm -1}$, 4 km$\,$s$^{\rm -1}$ for the v=2 line.

4. R Cnc (Fig. 10)

As for the other objects last included in our work (R Aql, $\chi$ Cyg, X Hya, and OH 26.5+0.6), the monitoring presented for this M6-M9 Mira variable is probably insufficient for concluding on the nature of its SiO variability. Nevertheless, the data presented here suggest that, also in this star, the SiO maser intensity follows the optical variations, with the typical phase lag of about 0.2 stellar cycles. The profiles of the two lines are not always similar, and the equivalent widths are $\sim$ 3.5 km$\,$s$^{\rm -1}$.

5. R Cas (Fig. 11)

This star has a large optical variability ( $\Delta m_v$ $\sim$ 6 mag) and the same holds for the SiO maser emission. The SiO flux curves in this star are typical of a Mira variable and show a contrast of $\sim$ 7. The v=2/v=1 integrated flux ratio has values between 0.5 and 1.0. The centroid of the v=2 line is, in general, placed at about 1 km$\,$s$^{\rm -1}$ to the red of the centroid of the v=1 line. The two lines have a similar shape, which presents strong changes during some SiO minima (see e.g. data at JD 2446450). The v=1 equivalent width increased from 3 km$\,$s$^{\rm -1}$ (JD 2446000) to 7 km$\,$s$^{\rm -1}$ (JD 2447300), dropped to 3 km$\,$s$^{\rm -1}$, and started to increase again. The mean values of the v=1 and v=2 equivalent widths are 4 and 4.3 km$\,$s$^{\rm -1}$ respectively.

6. o Cet (Fig. 12)

This star, also known as Mira, is the prototype of this variability class, which is named after it. Its optical variability reaches an amplitude of 6.5 mag with a period of 332 days. It belongs to a multiple system and it has been shown that the v=1, J=2-1 SiO maser emission is associated to the red giant (Hollis et al. 1990). The SiO curves of this object show a large contrast ($\sim~4-20$), the emission being undetectable during several minima, despite of that this object is in general a very strong SiO emitter. This fact, and its relative short period, makes this object the best suited for a detailed study on the SiO variability, in particular, to measure the phase lag of the SiO curves with respect to the optical one. Figure 12 clearly shows that this phase lag is of about 0.1-0.2 periods. The intensity of the maxima has been decreasing during the monitoring suggesting the possibility of a longer secondary period: however the verification of this point would require to extend the monitoring for several more years. The v=2/v=1 integrated flux ratio is, in general, between 0.5 and 1.5. During the monitoring, the line shapes of the two transitions have been similar except for a few short periods of time. The structure of the lines suffered important changes in several epochs, sometimes in coincidence with SiO minima (see data at JD 22446050, 22446750, and 22447400). The mean values of the v=1 and v=2 equivalent widths are 3 and 2.5 km$\,$s$^{\rm -1}$ respectively.

7. $\chi$ Cyg (Fig. 13)

This is the only S spectral type object in our sample. Due to the low abundance of Oxygen respect to Carbon (in comparison with M-type stars), the SiO masers of this type of objects are weaker than those of O-rich stars (no SiO masers have been detected in C-rich stars). The period of the optical variability of $\chi$ Cyg is similar to the 500 days of duration of our monitoring. Therefore, it is very difficult to establish whether the behavior of the SiO variability is regular or not. The previous works (Lane et al. 1982; Nyman 1985) do not help in clarifying this point. The v=2 maser is in general weaker than the v=1 one, and we report its detection only during a few epochs. The SiO masers showed strong changes in the line profiles during the monitoring of this source. The v=1 equivalent width is 3.7 km$\,$s$^{\rm -1}$, while for the v=2 line it is 2.7 km$\,$s$^{\rm -1}$ (on average).

8. U Her (Fig. 14)

The spectral type of this star is M7-M10 and the amplitude of its optical variability reaches values of about 5 mag. The observed SiO emission shows good examples of what we denominate "missing maxima'', i.e. epochs where an SiO maximum is expected but does not appear with enough contrast respect to the two adjacent minima. See, for example, the cases at JD 2446700 and 2447100. The mean contrast, due to these "missing maxima'', is low, $\sim$ 2.5. The v=2/v=1 integrated flux ratio does not have any regular variability and the mean value is $\sim$ 1. The centroid of the lines has a significant shift to the red between JD 2446700 and 2447100. In general, the changes of line shape are important from cycle to cycle. On average, the equivalent widths of the v=1 and v=2 SiO masers are 4.5 and 3.5 km$\,$s$^{\rm -1}$, respectively.

9. W Hya (Fig. 15)

This star is sometimes classified as a semi-regular (SRa), but the optical data we show in this work are not different from those of a typical Mira star: the variations are regular, with an amplitude of $\sim$ 4 mag. The SiO line shape is quite complex. The lines have always many spikes, the most intense ones seem to appear only after the optical maxima, with the typical phase lag of 0.1-0.2 (see for example the data around JD 2446900). The centroids of both lines remain within a range of 5 km$\,$s$^{\rm -1}$ during our monitoring; there is no systematic pattern in their variability. The v=1 and v=2 line shapes have important variations and, sometimes, the profiles at the two frequencies are quite different. Emission has been observed over a wide range of velocities (18 km$\,$s$^{\rm -1}$), the mean value of both equivalent widths being however moderate, 4.3 km$\,$s$^{\rm -1}$.

10. X Hya (Fig. 16)

In this object the signal to noise ratio is too low and the monitoring too short as to allow a detailed study of the properties of its SiO maser variability. Anyhow, the data suggest that the SiO curves follow the stellar cycle.

11. R Leo (Fig. 17)

This star is also a typical Mira variable, but the SiO variability is not very regular. There are some SiO maxima with a phase lag of $\sim$ 0.2 respect to the optical ones, but there is also a "missing maximum'' at the beginning of the monitoring. However, a Fourier analysis of the SiO variability curves yields a period that is consistent with the optical one. The two line centroids are very similar, presenting a secular shift of -4 km$\,$s$^{\rm -1}$ during the 6 years of our monitoring. The line shape changes frequently and, sometimes, the differences between the v=1 and v=2 maser profiles are important. On average, the equivalent width of the v=1 line is 3.9 km$\,$s$^{\rm -1}$; for the v=2 one it is 3.3 km$\,$s$^{\rm -1}$.

12. R LMi (Fig. 18)

This object is an M7-M10 Mira star with an optical variability amplitude of $\sim$ 5 mag. The regularity of the SiO data is good. The maxima are well defined for both v=1 and v=2 lines and the phase lag between the SiO emission and the optical is very clear. The mean SiO contrast is $\sim$ 3. The v=2/v=1 integrated flux ratio has a mean value of $\sim$ 0.7, except for the minima where it is $\sim$ 1. The centroids of the emission of the two lines change roughly in the same way. The mean value of the equivalent width of both lines is 3.6 km$\,$s$^{\rm -1}$.

13. IK Tau (Fig. 19)

This star shows regular optical variations with an amplitude of $\sim$ 4.5 mag, somewhat smaller than what is typical for Mira variables. It presents a large excess of infrared emission. The SiO data are very regular with a contrast of 3-4. The integrated fluxes of the two maser lines show very similar values over the different periods. In this case we present NIR observations of the star, from which it can be seen their excellent correlation with the SiO emission curves. The v=1 and v=2 line profiles appear very similar during the whole monitoring (two peaks, one at constant velocity and the other showing a secular shift of about 2 km$\,$s$^{\rm -1}$ in 1700 days). Due to the stability of the line profiles in IK Tau, we have detected SiO maser emission within a velocity range of only 10 km$\,$s$^{\rm -1}$. The mean values of the v=1 and v=2 equivalent widths have been 3.3 and 3.6 km$\,$s$^{\rm -1}$ respectively, ranging from 2 to 5 km$\,$s$^{\rm -1}$ during the monitoring presented here.

4. OH/IR objects

1. IRC +10011 (WX Psc; Fig. 20)

This is an infrared object, very probably a Mira-type variable with a very thick circumstellar envelope that makes very difficult to detect the central star at optical wavelengths. Herman & Habing (1985) have calculated a period of 632 days from a monitoring of the 1612 MHz OH maser emission in this object. Our SiO data seem to indicate a regular variation with a period of 500-600 days, in good agreement with the OH and IR periods. It is well known than in this type of objects both IR and OH maser emission vary practically in phase (see e.g. Habing 1996 and references therein). According to these data, an OH and IR maximum is predicted around JD 2446850. More recent NIR data from Le Bertre (1993) would indicate maxima at about JD 2446950 and JD 2447600. From our data we conclude that the SiO maser emission follows the NIR variations also in this source. The SiO contrast is moderate ($\sim$ 4). The v=2/v=1 integrated flux ratio shows a dependence with the phase, being $\sim$ 1 during the maxima and reaching $\sim$ 3 during the minima. The line profiles of the two transitions are very similar, with a strong change in shape around JD 2247500. The mean value of the equivalent width of both lines is 3.5 km$\,$s$^{\rm -1}$.

2. OH 26.5+0.6 (Fig. 21)

This object is also surrounded by a very thick dust envelope and its optical counterpart has not been detected. The 1612 MHz OH maser emission measurements by Herman & Habing (1985) indicate a period of 1566 days, predicting a maximum for JD $\sim$ 2447960, in good agreement with the SiO maximum at JD $\sim$ 2447950 found in our data. We cannot conclude at this point whether the SiO variability follows the IR curve or not, but we recall that usually the OH and IR curves are closely in phase. In general the v=2 line is in this object stronger and more complex than the v=1 line. The v=2/v=1 integrated flux ratio reaches a value of 8 during the minimum, and it is $\sim$ 2 for the maximum at JD $\sim$ 2447950. The v=2 equivalent width has a mean value of 4.5 km$\,$s$^{\rm -1}$, while the v=1 one is only 1.7 km$\,$s$^{\rm -1}$.

5. Young stellar objects

1. Orion IRc2 (Fig. 22)

This SiO maser is the only one in our monitoring that is not associated to an evolved star; the SiO emission in this region of massive star formation is associated with the infrared source IRc2 of the BN-KL nebula in the Orion A molecular cloud. There are only three known SiO maser sources associated to young objects in regions of very active star formation: Orion IRc2, Sgr B2(M) and W51-IRS2; the OH, H₂O, and CH₃OH masers being more common in these type of sources.

The line profiles of the v=1 and v=2 SiO masers observed in Orion IRc2 show two peaks (centered at $\sim-5$ and +16 km$\,$s$^{\rm -1}$) with almost no emission in the zone in between. This line shape has been essentially maintained since the discovery of these masers in Orion IRc2 in 1973 (Snyder & Buhl 1974). From VLBI observations (Greenhill & Moran 1986), it has been shown that these two spectral components are located in two different positions in the sky, separated by $\sim$ 0 $.\!\!^{\prime\prime}$14 (10¹⁵ cm for a distance to Orion A of 450-500 pc). The velocity centroid of the overall profile is about +5-+6 km$\,$s$^{\rm -1}$, close to that of the hot core surrounding IRc2.

The observed SiO variability does not show any regularity. The variations of the two components are independent, the blue one showing a larger amplitude. Also the variation of the v=2/v=1 flux ratio has no regularity. The two main components have several sub-peaks with lifetimes of a few months. The equivalent widths of the two components are very variable in the two transitions, with values between 3 and 7 km$\,$s$^{\rm -1}$.