The Massive Dust Reservoir in Supernova 1987A

Herschel image of the Large Magellanic Cloud showing the position of supernova 1987A

Figure 1 : A Herschel 250 micron and Spitzer 8 and 24 micron composite image of supernova 1987A and the surrounding regions. The SN is indicated by two horizontal lines.

Supernovae are very rare, and the closest one recorded in the last 300 years was detected in 1987, in a small galaxy close to the Milky Way. Using the European Space Agency's Herschel Space Observatory, UCL astronomers detected about 200,000 Earth masses of dust which has condensed out of the remains of the star which exploded. The dust grains contain the heavy elements which are so important for life, and the observations show that supernovae can be efficient dust-forming factories.


This discovery triggered great debate in the supernova community. The presence of so much dust is difficult to explain: it requires that the heavy elements in the debris of the supernova condensed into dust with virtually 100% efficiency. It also requires that most of the dust formed more than three years after the explosion, by which time many dust formation theories predict that it should no longer be able to form. UCL astronomers have followed up the original discovery paper, leading and contributing to several studies which have aimed to clarify when, where and how the dust formed.


The images from Herschel have quite a low spatial resolution, and so some astronomers argued that perhaps the dust detected was pre-existing material outside the expanding supernova remnant rather than new dust forming within it. To resolve this, observations with a much higher spatial resolution were obtained using the Atacama Large Millimeter Array (ALMA) in Chile. The ALMA images showed that the dust emission was indeed coming from within the remnant (Indebetouw et al., 2014; figure 2).

ALMA shows that far infrared emission is coming from within the expanding remnant of SN1987A

Figure 2. Observations with the Atacama Large Millimetre Array (ALMA) showed that new dust within the remnant rather than pre-existing dust elsewhere was the source of its far infrared emission.

Another question about the supernova remnant was whether the dust could have formed in the first few hundred days, but have been undetected. One way this could occur would be if the dust formed in very dense clumps. Dust in the centre of these clumps would be very cold and much harder to detect than strongly emitting warm dust. However, previous studies using radiative transfer models to reproduce the observed spectral energy distribution (SED) of the remnant found only relatively small quantities of dust at early times, and Wesson et al. 2015 reproduced these results. They found that by 1000 days after the explosion, the dust mass was strongly constrained to less than a thousandth of a solar mass, so that the increase of three orders of magnitude in mass must have occurred between 1000 and 8500 days after the explosion (figure 3). They also found that dust grains several microns in size -- far larger than grains found in the interstellar medium -- had to be present at late times to account for the observed SED.


Bevan and Barlow (2016) used the new DAMOCLES code to examine these findings with an independent method, based not on fitting the SED but on calculating the expected profiles of emission lines in the remnant, which are symmetric in the absence of dust but asymmetric if dust is present. Their results covered epochs from 714 to 3604 days after the explosion, and ruled out a large dust mass at those times. They also found that a dust size distribution of the type observed in the interstellar medium, and commonly assumed to exist in dusty astronomical objects, could not account for the observed line profiles.

Wesson et al. (2015) and Bevan & Barlow (2016) find a similar dust growth pattern using independent techniques to estimate the dust mass

Figure 3. The growth of the dust mass in SN1987A with time, as estimated from modelling the infrared emission (Wesson et al. 2015) and from optical line profiles (Bevan and Barlow 2016).

Page last modified on 11 apr 17 12:05 by Roger Wesson