Many observations indicate that accretion is highly time dependent, and is not the steady state process assumed by the most accepted magnetospheric accretion models. Stempels & Piskunov (2002) find that their veiling measurements vary with time, indicating a variable mass accretion rate. The time variations in line fluxes expected as accretion columns cross the line-of-sight, and excess continuum emission, are not always detected (Ardila & Basri 2000; Batalha et al. 2002). The observations of TW Hya by Alencar & Batalha (2002) show that line profiles vary substantially, despite the inclination of TW Hya being such that the system is seen nearly pole-on. Therefore the line profiles should not be affected by accre- tion columns rotating across the line-of-sight, or by a warped inner disc occulting the star (unless the large scale magnetosphere of TW Hya is tilted significantly with respect to the stellar rotation axis). However, the line profiles remain highly variable, suggesting that either the accretion rate is varying, or that the mag- netic field structure is changing significantly with time. Other IR observations appear to show evidence for warped inner discs, or accretion columns, periodically occulting the star (Johns-Krull & Valenti 2001).
Bouvier et al. (1999) present observations of AA Tau which exhibits deep photometric eclipses on a timescale equal to the stellar rotation period. Given the large inclination of AA Tau,i≈75◦
, the eclipses were attributed to a warped inner disc perhaps generated by a stellar magnetic field which is inclined to the stellar rotation axis. It was demonstrated by Terquem & Papaloizou (2000) that a tilted stellar dipole field interacting with a Keplerian accretion disc could indeed explain the deep eclipses observed in the light curve of AA Tau. M´enard et al. (2003), O’Sullivan et al. (2005) and Pinte & M´enard (2005) have since derived parameters for the geometry of the disc warp and the inner disc. Follow up ob- servations, combining medium term simultaneous photometric and spectroscopic monitoring of AA Tau, again indicated cyclic eclipses of the star by an inner disc warp (Bouvier et al. 2003). However, the warp appeared to disperse for one rotation period, only to return. During this missing eclipse both the line fluxes and veiling were strongly reduced and the variability was less, indicating that not only had the inner disc warp vanished, but at the same time the accretion process appeared to have switched off. There is evidence that this event may be part of a magnetospheric cycle, as I discuss below, where accretion occurs pe- riodically. This suggestion is supported by an observed correlation between the radial velocity of the redshifted absorption component (an accretion indicator) and the blueshifted absorption component (an outflow indicator) of the Hαline. Such anti-correlations between inflow and outflow signatures have been observed before, e.g. in SU Aur by Johns & Basri (1995). Bouvier et al. (2003) inter- pret this as direct evidence of the magnetospheric inflation model of Goodson & Winglee (1999). In this model the magnetosphere inflates over a timescale of several rotation periods, due to the differential rotation of field lines between the star and the disc, and is then disrupted, before returning to the initial field configuration. This interpretation is strengthened by more observations of AA Tau by Bouvier et al. (2007b), again combining long term photometric and high resolution spectroscopic data. There is evidence for variability on timescales of
days to months. They further demonstrate that during the deep photometric eclipses, attributed to the warped inner disc, the veiling and HeI 5876˚A line flux reach a maximum. At the same time large redshifted absorption components appear in the Balmer lines of hydrogen. Putting the pieces of evidence together this suggests that the accretion shock, the accretion column and the inner disc warp are at the same rotational phase. Furthermore, Bouvier et al. (2007b) find that the mass accretion rate varies over several rotation periods, with evidence for strong accretion rates coinciding with a weak hot wind, and weak accretion rates when a strong hot wind is observed. Again this has been interpreted as evidence for magnetospheric inflation, although it should be stressed, that this has only been observed on one star, and more long term monitoring campaigns are required (e.g. the initial results presented by Grankin et al. 2007). It is also worth noting that the dataset for AA Tau now spans a decade and it has been found that as well as the short term variability attributed to magnetospheric in- flation, there is tentative evidence for a magnetic cycle. Although deep eclipses have been observed consistently for many years on a timescale of AA Tau’s ro- tation period, the shape and depth of such eclipses have changed over the years, perhaps indicating the presence of a magnetic cycle (Bouvier et al. 2007b).
The observed correlation in AA Tau between signatures of accretion and out- flow has been interpreted as evidence for magnetospheric inflation. This refers to a class of models which predict cyclic changes in the structure of the magneto- sphere with alternating episodes of accretion and outflow (Aly & Kuijpers 1990; Bardou & Heyvaerts 1996; Goodson et al. 1997; Goodson, B¨ohm & Winglee 1999; Goodson & Winglee 1999; Agapitou & Papaloizou 2000; Matt et al. 2002). The basis for such time dependent models is that the differential rotation between the foot points of field lines connecting the star to the disc, causes the field lines to expand (see Fig. 1.7). After a few rotation periods the twisted field lines are torn open, ejecting material from the system. The open field lines then recon-
Figure 1.7: An illustration of magnetospheric inflation, reproduced from Good- son & Winglee (1999). The solid black lines are field lines that constitute the original stellar magnetosphere. Such a model is believed to explain the observed correlation between accretion and outflow signatures in AA Tau (Bouvier et al. 2003, 2007b). Reproduced with permission.
nect, uploading disc material onto the stellar magnetosphere (Hayashi, Shibata & Matsumoto 1996; Romanova et al. 2002). After the accretion episode the initial field configuration is restored. In such a way the accretion process is periodic, and intimately connected to ejection processes. There are further observations which suggest that the magnetosphere is dynamic and is strongly influenced via the interaction with the disc. Oliveira et al. (2000) present observations of SU Aur which indicate that field lines are being twisted by differential rotation. Alencar, Johns-Krull & Basri (2001) and Ardila et al. (2002) find evidence for reconnec- tion events, with periodic high velocity outbursts identified via the blueshifted absorption components of emission line profiles.
Highly time variable accretion is a feature of other accretion models as well (e.g. Miller & Stone 1997), however, the most comprehensive numerical modelling of the star-disc interaction are the 3D simulations being carried out as part of the US-Russian collaboration on plasma physics. Koldoba et al. (2002b) and Romanova et al. (2003) demonstrated that the interaction between the inner disc and an inclined dipolar magnetosphere often leads to the creation of a disc warp as well as another secondary warp created by the tendency of matter within the magnetospheric cavity to corotate with the stellar magnetosphere. They predict the existence of many accretion columns connecting the star to the disc, whose geometry changes in time. They argue that such accreting structures may be responsible for the large variability seen in accreting T Tauri stars. This work was extended to determine the distribution and shapes of accretion hot spots and the resulting photometric variability (Romanova et al. 2004a). Furthermore, they have also considered field geometries which are more complex than a dipole (Long et al. 2007); however their initial field configurations are still axisymmetric. Long et al. (2007) do find that the location and distribution of hot spots is strongly dependent upon the geometry of the magnetic field; this is also a prediction of my complex field accretion model discussed in Chapter 4, and published in Gregory
et al. (2005, 2006a).
Many accretion models have been developed that attempt to combine accre- tion with ejection processes (stellar winds, disc winds, X-winds and bipolar jets). The Shu model, discussed in the previous section, was one of the first models to attempt to combine accretion with disc winds and jets. In the Shu model an ac- cretion powered disc wind, originating from a small region close to the corotation radius, carries away angular momentum from accreting material, preventing the star from spinning up (Shu et al. 1994a; Ostriker & Shu 1995). Various other models have been proposed studying the connections between accretion and ejec- tion processes, which tend to differ on exactly how the field interacts with the disc. Goodson et al. (1997) show that field lines threading the disc can drive winds and jets, whilst Hirose et al. (1997) argue that reconnection events can drive jets and explain the slow rotation of CTTs. Such accretion-ejection models often provide very efficient spin down mechanisms (e.g. Ferreira et al. 2000; Romanova et al. 2005). One of the most efficient such models is the Reconnection X-wind model (Ferreira et al. 2000), which is compared to other outflow models and their ob- servable predictions by Ferreira et al. (2006). In this model a dynamo generated stellar magnetosphere reconnects with the remnant disc field, deflecting the disc material away from the disc plane. A fraction of the material is accreted with the rest emerging as an MHD jet, the reconnection X-wind. Such outflows are fed with disc material, but are powered by the stellar rotational energy. Thus unlike the Shu X-wind model, the reconnection X-wind provides a torque that brakes the star efficiently.
An alternative accretion model is being developed which does not assume that material is mass loaded onto field lines (von Rekowski & Brandenburg 2004a; von Rekowski et al. 2004b; von Rekowski & Brandenburg 2006a; von Rekowski & Piskunov 2006b). The basis of this accretion model is that the both the stellar and the disc magnetic fields are dynamo generated and maintained. Material
accreting from the disc is in constant competition with the outflowing stellar wind from the star. Accretion is episodic, with material only able to accrete onto the star into regions where, and at times when, it is able to overcome the stellar wind (von Rekowski & Piskunov 2006b). There is of course a lot of evidence that accretion is a highly time dependent process, however, in this model accretion only occurs onto the star’s equator; it therefore remains to be seen if such a model can explain the level of photometric variability which is often attributed to the presence of hot spots at higher latitudes. Furthermore, von Rekowski & Piskunov (2006b) argue that although material may reach several hundred kms−1 in the
gap between the star and the inner disc edge (consistent with the large redshifted absorption components of inverse in P-Cygni profiles), it slows to only a few tens of kms−1 before accreting onto the star. This raises the question as to whether
hot spots can be explained at all by this model. Of course, this model, as with the many others which are attempting to explain the deep connections between time dependent accretion and outflows processes, will continue to develop in the future. Reconciling the wealth of observational data of the highly variable and complex CTTs systems is truly a difficult task, however, much progress is likely to be made over the coming years.