3.5. The Effects of Multiplicity

The evolution of circumstellar disks can be dictated by either internal (e.g., viscous accretion, gravitational instability, planet formation) or external processes. Examples of the latter include ultraviolet photoevaporation in the vicinity of a massive star (e.g., Johnstone, Hollenbach, & Bally 1998), dynamical interactions with other stars in a crowded cluster environment (Kroupa 1995; Boffin et al. 1998) or in a local multiple star system. In the low stellar density Taurus-Auriga region, which is devoid of stars earlier than A0, the dominant external process affecting disk evolution is expected to be dynamical star-disk or disk-disk interactions in multiple star systems. Most young stars in nearby clusters and main-sequence stars in the field are in multiple systems, and the multiplicity fraction in Taurus-Auriga may be exceptionally large (Mathieu 1994; Mathieu et al. 2000). The similar multiplicity fractions for YSOs and main-sequence field stars indicates that binary formation occurs early in stellar evolution (at least before the Class II stage, and likely much earlier), when significant circumstellar material is still present. Gravitational interactions are expected to severely affect the structural integrity of disks in the system, including truncation of the outer parts of individual circumstellar disks, gap formation in circumbinary disks, or even complete dissipation of circumstellar material via accretion or ejection (Artymowicz & Lubow 1994).

As an example, consider a young binary system with semimajor axis a and eccentricity e, which also harbors two individual circumstellar disks and a larger circumbinary disk. Simulations of the gravitational dynamics in such a system indicate that the circumstellar disks will be truncated and a gap will open in the circumbinary disk, at radii which are determined primarily by the values of a and e (Artymowicz & Lubow 1994). The disk model described in §3.1 can be adjusted to determine the effects on the SED of such disk configurations by re-setting the inner and outer radii for the various disk components or simply by setting {tex}\Sigma_r = 0{/tex} for the cleared regions (e.g., Jensen, Mathieu, & Fuller 1996). One expected result from these SED models is that the submillimeter emission should be significantly diminished for systems with a projected semimajor axis ({tex}a_p{/tex}) on the order of a few tens of AU, but essentially identical to single stars for small and large {tex}a_p{/tex}.  Previous observations have indicated that {tex}a_p \frac{<}{~} 50-100 AU{/tex} binaries have less submillimeter emission than single stars or wider binaries (Jensen, Mathieu, & Fuller 1994, 1996; Osterloh & Beckwith 1995), with the important exception of spectroscopic binaries ({tex}a \frac{<}{~} 1 AU{/tex}; e.g., Mathieu et al. 1995, 1997).

Table 3 gives a list of multiple stars in the sample and their projected separations. This information and the data in Table 1 have been compiled in Figure 11, which shows the 850 µm flux density and disk mass as a function of projected semimajor axis. To be consistent with previous work, spectroscopic binaries as well as Class I sources have been excluded in this figure and the analysis that follows. Unresolved higher-order multiple star systems ({tex}\geq 3{/tex} stars) without resolved observations in the literature were assigned the same {tex}F_v{/tex} or {tex}M_d{/tex} value for all projected separations. Notes on assigning values for a few other systems are provided in Table 3. We utilize a variety of two-sample statistical tests that incorporate upper limits to determine the probabilities that the 850 µm flux densities and disk masses in various binary subsamples are drawn from different parent distributions. The total sample is separated into categories based on projected semimajor axis, resulting in three groups: close binaries with separations less than some critical value, {tex}a_c{/tex};  wide binaries with separations larger than {tex}a_c{/tex}; and single stars.

Table 4 lists the ranges of probabilities that the various subsamples for {tex}a_c{/tex} = 50 AU and 100 AU differ from a sequence of survival analysis statistical tests performed with the ASURV software: the logrank, Peto & Peto, Peto & Prentice, and Gehan tests (see the detailed descriptions by Feigelson & Nelson 1985). The same tests were also performed for Class II objects only. The results in Figure 11 and Table 4 confirm the earlier conclusions of Jensen, Mathieu, & Fuller (1994, 1996): submillimeter flux densities and disk masses are significantly lower in close binaries ({tex}a_p\leq 50-100 AU{/tex}) than wider or isolated systems and wide binaries essentially have the same disk masses as single stars. These differences are greatest for a critical semimajor axis {tex}a_c = 100AU{/tex}.  The results for the total sample (i.e., when Class III binaries are included) generally exhibit lower probabilities than the subsample of only Class II objects in Table 4, with the exception of the close and wide binary populations with the {tex}a_c=100AU{/tex} cutoff criterion. These differences are likely due to the evolutionary behavior of disks between the Class II and III stages (see §4), rather than an environmental effect in the multiple system. The exact probabilities for the various subsamples appear to be fairly sensitive to the assignment of flux densities or disk masses in unresolved higherorder multiple systems. High resolution interferometric observations are needed to determine the relative submillimeter contributions of individual components in these systems.

In a statistical sense, the presence of a companion in the range of ~1100AU decreases the apparent disk mass(es) in the system, presumably due to enhanced accretion onto the stars and/or dispersal into the local ISM. However, multiple star systems still contain disks, as evidenced by the relatively high detection rate in the submillimeter, 66 ± 10% for multiple systems compared to 58 ± 8% for single stars, as well as other inner disk signatures in the optical and infrared (e.g., White & Ghez 2001). High-resolution interferometric measurements of disks in multiple systems have revealed a number of important exceptions to the statistical analysis above. For example, the GG Tau A and UZ Tau systems both have small projected separations but very large disk masses, the former in a circumbinary disk and the latter in a pair of disks with four stellar components (Koerner, Sargent, & Beckwith 1993a; Dutrey, Guilloteau, & Simon 1994; Jensen, Koerner, & Mathieu 1996). Moreover, single-dish continuum surveys may be missing signatures of outer disks in close multiple systems: a close binary ({tex}a_p = 32AU{/tex}) in the SR 24 triple system in Ophiuchus was found to have a large circumbinary gas disk detected in CO line emission but not in the continuum due to its low mass (Andrews & Williams 2005). The question of whether the large fraction of young stars in multiple systems could eventually harbor planetary systems will remain unanswered until more detailed case studies with interferometers (e.g., Jensen & Akeson 2003) can confirm the properties of their disks.