Extrinsic Triggers for Nova Formation in the Andromeda Galaxy
By Alex H., Margaret W., and Julie K.

This paper has been published in the 2003 RBSE Journal. Click here to view in .PDF format

Introduction

For the past five years the students in our high school have been collaborating with astronomers from the National Optical Astronomy Observatory in Tucson, Arizona. Through the Research-Based Science Education program (RBSE), students are supplied with CCD images of the Andromeda Galaxy and specialized image-processing software to search for novae. It is then up to their discretion regarding what types of investigations to develop.

Novae are a type of cataclysmic variable star that flare brightly when the white dwarf of a close binary pair gains mass from the other star. The nova’s apparent magnitude increases by a factor of 10,000 to 1,000,000, lasting for only a short period of time. The nova flare is caused when hydrogen from the red giant of the nova binary is accreted into a swirling disk of matter surrounding the white dwarf. When a minimum of .001 solar masses of hydrogen is accreted, the hydrogen burns explosively, causing the white dwarf to flare.

Upon studying the accumulated Nova Search data, it seemed that novae outbursts tended to form in close proximity to each other. This had many students searching for the space and/or time relationship for the novae. If a student successfully discovered a relationship, the question arose: why would the novae be related? It has been beleived that only the partner star of the white dwarf can cause novae formation by means of accretion. However, the mass accreted onto the surface of the white dwarf could possibly be the result of an extrinsic mechanism.

Purpose

The aim of this investigation is two-fold. It is to search for patterns in the distribution of novae, either in space or time. With these patterns, a question arises which needs to be addressed. Could nova outbursts be the result of an extrinsic mechanism, such as the shell of a planetary nebula or the tidal force of a globular cluster?

Procedure

The first portion of the procedure involved determining if the novae discovered from the high school nova search project had any relation to each other in space and time. The data consist of 27 sets of CCD images taken over a period of six years by astronomers and teacher researchers at Kitt Peak National Observatory. The CCD images span a diameter of the central 32,000 light years of M31, and cover a total area of 4.8x10⁸ LY. The images were taken using the Kitt Peak 0.9m and 2.1m telescopes and a Hydrogen-Alpha filter. The H-alpha filter has a passband of 650-660 nm. Since novae emit strongly in these wavelengths, they can be separated in the images from other types of variable stars.

The CCD images are divided into subdivisions called subrasters in order to make them a manageable size to work with on a typical school computer. 21 epochs (nights) of data were analyzed covering a time range from September 1995 to January 2002.  The first goal of the research was to examine a possible space and/or time relationship between the novae.  Detailed maps of each subraster with the novae found in every epoch were created.

The image analysis was done using a scientific image-processing program called Scion Image and specialized macros developed by NOAO to do photometry on FITS images. When searching for new novae, students confirmed the discovery of a nova with two other students, using a double-blind method to ensure that the nova was real.

An optical image of M31 with Nova Search images overlaid, and novae marked in red

Space and Time Relationships

Using Adobe Photoshop 6.0, the locations of all the novae were plotted onto an epoch mosaic. Then, all of the marked mosaics were exported into Scion Image, where they were blinked, or animated, in order to find any type of pattern in time. Finally a mosaic composite of all subrasters of data was created and the image was analyzed for any spatial novae patterns.

The next step in finding any type of spatial pattern was to create an optical image of M31, then overlay the plots of novae onto this optical image. Using The Atlas of the Andromeda Galaxy[1], images were gathered of the area near the core being studied. These images were then cropped and fitted together in Adobe Photoshop to form one large image mosaic. This picture was then cropped further and enhanced to match the image size to the actual size of the Nova Search parameters. The plots of all novae were then imported into Adobe Photoshop and merged on top of the optical image. In all, 394 novae were plotted. Novae located in the core area, where features could not be ascertained, could not be examined. The novae located around the central area were then examined for further patterns. This reduced the original number by 249. The novae located around the central area were then examined to see if their locations correlated with dust clouds, stellar associations, or arm structure.

Extrinsic Mechanisms

For the second goal of this research, a list of possible extrinsic triggering mechanisms for novae was developed. Ideas for possible mechanisms were supernova remnants, planetary nebulae, globular clusters, x-ray sources, and other novae.

Supernovae were an obvious choice, due to prior knowledge of their effect on the formation of stars. During a supernova, 80% of the star’s mass is released into space. It has been documented that supernovae can cause star formation.[2] As witnessed by SN 1987a, shockwaves from the supernova trigger regions of formation in the surrounding interstellar gas. This raises the question, if a supernova causes star formation, then what other stellar events could it induce? More specifically, can a supernova induce nova formation?

Planetary nebulae are stars that in the late stages of evolution release shells of gas. As the gas from the outer part of the star is released, the core of the star heats up and eventually reveals itself as a white dwarf. When the shell is completely released, the white dwarf heats up the gases and causes them to glow. Much like a supernova, the mass is released into space and could possibly accrete onto the white dwarf.

Globular clusters provide an interesting aspect to the search for a novae trigger. Globular clusters tend to have an orbit that goes through the plane of the galaxy, rather than orbiting along it. They also tend to be fairly massive objects in comparison with the novae. This abundance of mass could potentially aid the accretion of hydrogen from the red star onto the white dwarf, by creating a massive tidal effect and helping to induce the mass transfer between the partners.

Another characteristic of globular clusters is their tendency to travel through the plane of the galaxy.  Due to this unusual orbit, globular clusters like Palomar 5 undergo a process called tidal shredding.  The tidal forces of the galaxy tears stars away from the globular cluster leaving a trail of stars behind and in front of the cluster that can be 6000 light years long.  In the case of Palomar 5, the tails contain nearly 1.3 times the mass of the globular cluster itself. These tails may interact with the binary system, aiding in mass accretion.

X-Ray sources from the active nucleus of the galaxy could also participate in the formation of novae. The gases that escape from the active areas in the nucleus are in a hot cloud of wind. These massive clouds could possibly aid the accretion of the white dwarf.

Pulsars are rotating neutron stars that release “pulses” of waves at incredibly stable intervals. Their polar areas emit jets of materials in a manner similar to X-rays.

Once this list was narrowed and defined, a literature search was performed to find outside information on the possible sources of novae. The Nova Rate in the Elliptical Component of NGC 5128[3] provided an idea linking globular clusters to novae flares. SIMBAD[4]  and VizieR[5] provided the most of the information about the locations and types of objects in M31. VizieR allows the user complete access to extensive astronomical catalogues and data tables.  The Atlas of the Andromeda Galaxy and Regulus![6] provided detailed optical images and maps of the Andromeda Galaxy.

In addition to the online databases, astronomer George Jacoby from NOAO provided a list of planetary nebulae in M31 in response to an email request.

Controls and Limits

The parameters of the field of study were determined by utilizing the coordinate feature in Scion Image. This feature allows the user to drag the cursor over the point and read its right ascension and declination. After the limits of the available Nova Search data were determined, their coordinates became the indicator as to whether or not the data found in VizieR was useable. This decision to narrow the study to only the central region of M31 was twofold; the novae indicated from Nova Search were discovered by students at our school. Because of this, the approximate time of outburst was known.  The other sets of novae were discovered over much longer periods of time and had no record of when the outbursts occurred.

The second reason this field was selected is that it represents only one population of stars. The stars in M31, much like the stars of other galaxies, are of differing ages. The stars located in the bulge are considered to be Population II stars, or old stars. By limiting the field of data to the central regions of the galaxy, it was then possible to study only one stellar population.

Data sets were acquired for the location of novae, supernova remnant (SNR) candidates, X-ray sources, planetary nebulae, and globular clusters.  No pulsar data were available for M31.

Data Processing

After determining which data fit the parameters of the Nova Search information, the data was imported into Microsoft Excel to be converted to a useable form for PSI Plot. Right ascension is measured in hours, minutes, and seconds while declination is measured in degrees, minutes, and seconds. Although there is a time feature in Excel, it does not convert this format properly, so it was necessary to write a program that would change the minutes and seconds of both right ascension and declination to decimal interpretations. By hand, changing minutes and seconds is an easy task, however, with the enormous amount of data acquired, the following conversion functions were created in Excel:

RA =MID(A4,3,2)/60+MID(A4,6,5)/3600
DEC =MID(A5,1,2)+MID(A5,4,2)/60+MID(A5,7,5)/3600

These numbers were then entered in PSI Plot. Graphing in PSI Plot allows for the user to indicate symbol size, symbol type, and colors. Using the XY Plot feature, graphs were created to analyze the spatial relationship between novae and their possible triggers. All novae were given a specific symbol and separated by color according to whether they were found through Nova Search or on VizieR. The other objects were plotted against the set of novae and finally all the PSI graphs were plotted against the optical pictures of the Andromeda galaxy.

Modeling Mechanisms for Extrinsic Triggers

Upon research, it was discovered that, in order for a classical nova to flare, it must have .001 solar masses. While this may seem like a small number, it was crucial to the determination of the plausibility of several of the triggering mechanisms. Simple math constructs were created to determine the “sphere of influence” each mechanism might exert upon the nova. The constructs were based on the average values for the physical and dynamic characteristics of each object or phenomenon. The averages were then used in calculations to determine the possibility of influence on the novae.

Average Reference Values for Selected Objects

Globular Clusters: Mechanism of Action

Using Newton’s law of universal Gravitation, mathematical models were created for the interaction between the red giant - white dwarf binary star pair (the nova pair) and the binary star pair with a globular cluster. Using the average mass for red giants and white dwarfs, it was determined that the average force needed to accrete mass from the red giant to the white dwarf was around 2.56x10³⁸ N. The maximum distance that a white dwarf must be from its red giant partner is 7.4x10¹¹ meters.

Force = G(average mass red giant(kg))(average mass white dwarf(kg))
(maximum distance between binary pair (m))²

2.56x10³⁸N = (6.67x10^-11 kg) (1.79x10³⁰ kg)(1.50x10³⁰ kg)
(7.0x10¹¹m)²

Force = G(average mass globular cluster(kg))(average mass binary pair (kg))
(maximum distance between binary pair and globular cluster core)²

2.56x10³⁸N = (6.67x10^-11 kg)(3.80x10³⁵ kg)(3.29x10³⁰ kg)
   (3.07x10^-18m)²

This equation determined that a nova could flare only if it were located near the center of the globular cluster. The results indicate it would be impossible to create a nova based purely on the tidal force from a globular cluster.

An article written by Robin Ciardullo et al[7], suggests a different type of triggering mechanism by which globular clusters could form novae.  Theoretically, the novae form within the globular clusters by cluster core collapse and are then ejected after they are formed.  The ejection process has been hypothesized to be the result of three-body interaction or tidal disruptions.

If the Ciardullo hypothesis is correct, the density of the novae should be directly correlated to the density of the globular clusters. Essentially, the higher the density of globular clusters, the higher the density of novae.

Supernovae: Mechanism of Action

In the investigation of the possibility of the supernova effect, another construct was developed to examine the mathematical likelihood of such a trigger. In order to determine the efficacy of such a system, it was necessary determine the area and volumes of several types of stars and their influences.

The average white dwarf was found to have a surface area of 5.14x10¹⁴m². However, the surface area that would be affected by a supernova was estimated to be around 50% of the total surface area of a white dwarf, having 2.57x10¹⁴m² being exposed to the supernova shell. In order for the nova to flare, the mass accreted onto the white dwarf from the supernova would have to be greater than or equal to 1.99x10²⁷ kg, or .001 solar masses.

The average radius of the progenitor star is 3.49x10⁹m.  In a supernova, 80% of the progenitor’s radius becomes the thickness of the original supernova’s shell.  This thickness was calculated to be 2.79x10⁹m.  In order to determine the kilograms of mass accreted onto the white dwarf by the supernova, half the surface area of the white dwarf multiplied by the thickness of the shell determined the kilograms accreted onto the white dwarf.

(.5 surface area white dwarf)(thickness shell)(density progenitor) = kg accreted onto white dwarf

(2.57x10⁹m²)(2.79x10⁹m)(4.43x10³ kg/m³)=3.18x10²⁷kg

The inverse square relationship was utilized to calculate the maximum distance allowed for a supernova to cause a nova to flare.

(distance)= √[(mass accreted by SN shell)/(mass required for nova to flare)]

1.26 m = √[(3.18x10²⁷kg)/(1.99x10²⁷ kg)]

Using the universal law of gravitation it was determined that supernovae were not plausible candidates as extrinsic triggers for novae outbursts.  Their required proximity to influence a nova was actually closer to the nova than its partner star.  Supernovae were still included in the data because they had the possibility of offering clues into the locations of stellar associations.  However, there were so few supernovae and supernova remnants in the field of study that it was impossible to discern any meaningful relationship between SNRs and stellar association.

In realizing the mathematical impossibilities of supernova remnants as an extrinsic mechanism, it was then determined that planetary nebulae would also not function as a trigger for novae due to their lower masses.  Planetary nebulae would theoretically function in the same manner as supernovae, but to a lesser degree in its “sphere of influence”.  However, planetary nebulae were kept with the database because they helped define the central area of the galactic bulge.

X-rays were examined to due to their unique structures. A normal X-ray source, such as a black hole or pulsar, usually forms a disk of X-rays. At the poles of the source, however, material is blasted into space in a long jet of matter. It is possible that this material would be directed at a white dwarf star, and that the material being blown out by the X-ray source would aid in the accretion of the white dwarf, therefore causing a nova. 

Controls and Sources of Error

When using VizieR to find supernovae, data was only used if it had a high or medium confidence instead of low to insure that the supernovae in question actually existed. To find the limits of Scion Image the mouse was moved pixel by pixel to determine the change in right ascension and declination. It was determined to change ±.7 seconds with the movement of one pixel in all directions.

The Andromeda Galaxy is a roughly circular shape but is inclined 13 degrees to the line of sight. Because of this, the galaxy appears to be more elliptical to us. This was not compensated for in the plots, but it was assumed that the parallax effect it caused was small in comparison to the distance to M31 (2.15 million LY.)

All of the graphs made were plotted using a two dimensional XY plot and plotting RA and declination. There was no data for the radial distances to the objects. As a result, objects that appear to overlap may, in reality, be separated by an extreme distance of thousands of light years.

Sources of data:

Planetary Nebulae – George Jacoby

High School Novae – RSBE

VizieR Gathered Novae – Combined General Catalogue of Variable Stars (Kholopov+ 1998)

Globular Clusters – http://cfa-www.harvard.edu/~pbarmby/m31gc/m31gc.coo

Supernova Discovered by Sources Outside VizieR - New SNR Candidates in M31 (Magnier+ 1995)

High Confidence Supernovae – New SNR Candidates in M31 (Magnier+ 1995)

Medium Confidence Supernovae – New SNR Candidates in M31 (Magnier+ 1995)

1885 nova – Combined General Catalogue of Variable Stars (Kholopov+ 1998)

X-Ray – Einstein IPC data 1998, ROSAT, Zim 2001, XMM Newton Observatory 2002, Chandra 2003

Analysis

When the total mosaic of all novae discovered from the school Nova Search was examined, three interesting patterns emerged. The first was around the southwest of the galactic core. Several novae appear in a ring 8000 LY long that seems to wrap around and extend away from the core of the galaxy. While at first it may seem that the pattern is only due to its close proximity to the core, other novae exist even closer to the center of the galaxy. Therefore, the ring does not necessarily encompass the core itself.

The second pattern is located to the east of the core. A long string of novae stretches across 2 subraster images, creating a distinct chain roughly 5000 LY long and 1600 LY wide. The third pattern is located southwest of the core. Two diagonal lines of novae can be traced.

All GP novae plotted against the optical image. The three patterns can be seen in this picture.

There are two main ideas that may explain these occurrences. One explanation is that the novae comprising the ring and the east chain are spiral arm structures that are extending from the core. This would explain the wrapping pattern the ring has on the central area of the galaxy. The other explanation is far more intriguing. An article written by M. Irwin et al[8] suggests that the tidal force of M31 is actually pulling matter from its two satellite galaxies, M32 and NGC205. It is possible that these arcs mark material that is being transferred from the dwarf galaxies to M31, and that the greater occurrence of novae is due to a greater amount of matter accreting on the surface of a white dwarf star. Alternatively, the novae may mark a population of stars that is being transferred into M31 by merger.

A proposed inflow path of the tidal merger of stars from NGC 205

The images created with the Atlas of the Andromeda Galaxy showed the locations of various dust clouds and stellar associations. When the image was examined, it appeared that novae were clustered in and around dust clouds and stellar associations. Of the total 145 novae left after the first reduction, 60 were found near or in the clouds and associations, a total of 41%. Out of the 60 of the novae found around these objects, 13.3% were found on the edge of stellar associations, 15% were found in stellar associations, 41.6% were found on the edge of dust clouds, and 30% were found in dust clouds. Overall, a greater percentage of novae occurred in dust clouds rather than stellar associations, but 59% of all novae were not associated with either feature.

In an online literature search, it was discovered that the center of M31 has a double nucleus.  However, it was determined to be only 30 light years across whilst the data field spanned 24,000 light years.  The size of the double nucleus compared with the size of the data was not large enough to determine any significant influence. 

Extrinsic Mechanisms - Analysis

North Novae The novae are more numerous in the center, although there are a considerable number of novae in the upper left corner. The most interesting is the cluster in the east center of the graph. Another interesting aspect is what has come to be known as "the finger" in the lower center of the graph. The region moves down from east to west and no novae are present in it. This is most likely because something in the galaxy has obstructed the view.

Other Novae The most obvious pattern is there are more novae closer together in the middle of the graph. This is most likely because this is where the center of the galaxy is located. The finger region appears clearly in this graph as well.

North Novae vs. Other Novae
The North novae follows the same pattern as the other novae. There are no novae at all in the finger region.

North Novae vs. All Novae. The finger region is located at approximately between 41.2 and 41.3 declination, .715 and .710 Right Ascension

All Novae There are more novae present in the center with a rather large clump of 5+ novae in the left center region. The finger is very obvious.

Planetary Nebulae Like the novae, the planetary nebulae are also more frequent in the center because this is the center of the galaxy. The finger region is also evident in this graph, but it is not as pronounced. In the upper right and lower left sides of the graph there are very few points. This is most likely because the plane of the galaxy is where most of the planetary nebulae occur.

Planetary Nebulae vs. All Novae  Some planetary nebulae appear in the finger region. Distribution of the planetary nebulae is more concentrated toward the galactic center than the distribution of the novae. This leads to the idea that a number of novae observed may be in the halo instead of in the disk of the galaxy.

All Supernovae and Remnants The supernovae and remnants were very spread out overall, with no clumping in the middle, as is seen in the other graphs. There is no apparent pattern.

All Supernovae and Remnants vs. All Novae None of the supernovae and novae overlap, or are even very close to each other, unlike the comparisons of novae with the planetary nebulae and the globular clusters. There are no real patterns evident between the novae and the supernovae.

Globular Clusters  Globular clusters are more random in their distribution than the novae and planetary nebulae. The globular clusters do not outline the shape of the galactic plane.

Two of the “glob” novae occur at the same location, separated by an interval of two years.  The second outburst lasted less than a week, based on the examination of other epochs.  This is intriguing because if it represents a recurrent nova, the period is much shorter than what has been previously documented by other observers.  Typically, recurrent novae have a dormant period of decades.

A possibility exists that the two novae or in fact separated in space by a radial distance of light years.  The star may also represent a type of dwarf nova with a short outburst period, such as an SU Ursa Majoris variable.[9]

All Novae vs. Globular Clusters There is a small cluster of five novae that are clumped together in subraster 10 of the school Nova Search data. However, attempts to overlay the “glob” onto any other optical images failed, due to the fact that the cluster appeared in the overexposed area of galaxy. Therefore, no explanation for this clustering is readily apparent.

As the density of the globular clusters increases, so does the density of the novae. This supports the hypothesis that globulars may form some novae. The mechanism may be core collapse or perhaps accelerated accretion by tidal tails of shredded globulars such as Palomar 5.

The graph of all novae vs. globular clusters. Note the absence of nova in the so-called "finger area"

X-Ray The X-ray sources are more scattered, unlike most of the graphs which have the majority of their points located in the center. The most intriguing part of this graph is the line of points in the center.

X-Ray vs. All Novae  It appears that many of the novae and X-rays overlap, but when magnifying the graph results show that this is not true. There are no real patterns between X-Rays and Novae.

The Finger

The finger is a region that is visible in almost all the graphs that contains no novae and very few of the other objects. The reason for this may be a dust cloud located in the galaxy where the finger is on the graph. The finger is 8000 LY long and 1600 LY across. The dust cloud is obscuring the picture, so only foreground objects can be seen.

Conclusions

No pattern in time for novae flares is apparent. However, several patterns in space exist, recorded over a relatively short time span of six years. These may possibly related to the spiral arm structure of Andromeda or jets of material streaming from NGC 205 and M32 into M31.

Nova density increases closer to the core of the galaxy, due to greater concentration of stars near the center of the galaxy. As the density of novae increases, the density of globular clusters increases. This supports the idea that globular clusters may aid in the formation of novae. This may be either as the result of cluster core collapse or the increased accretion due to the tidally shredded tails.

The discovery of one nova, which appears in epoch 3 and again in epoch 19, may be a newly discovered type of recurrent nova which has a period of two years.

There are no apparent correlations between novae and their locations relative to dust clouds, stellar associations, supernovae, planetary nebulae or X-ray sources.

When the “finger” area was investigated, optical pictures showed that this lane of the galaxy appears obscured by dust, therefore reducing the visibility of novae.

Extensions

            There are several directions this research could turn in the future. Since it has been proposed that there is a possible correlation between novae and globular clusters, research needs to be done on whether this is caused by a tidal streamers due to globular shredding, or by core cluster collapse. The prospect of the dwarf galaxies NCG 205 and M32 causing novae formation by tidal streaming is also an interesting area that could be further investigated. Finally, the single recurring nova found does not seem to fit any of the other forms of recurring novae. This suggests that a new type of recurring novae may have been discovered.

Acknowledgements

            We would like to thank George Jacoby for providing planetary nebulae data and insight into nova formation.  We would also like to thank Connie Walker for aiding us in determining some of the mathematical models, Andrew F. for helping us create an Excel program to convert the coordinates, and Paul F. for providing suggestions in calculating the areas of our plots.  Finally we would like to give our great appreciation to our teacher for guiding the course of our research and providing great insight and leadership throughout.
 

[1] http://nedwww.ipac.caltech.edu/level5/ANDROMEDA_Atlas/Hodge_contents.html

[2] http://www.telescope.org/pparc/res8.html

[3]Ciardullo, Robin, et al. “The Nova Rate in the Elliptical Component of NGC5128.” The Astronomical Journal.
     April 1990.

[4] http://simbad.u-strasbg.fr/Simbad

[5] http://vizier.u-strasbg.fr/viz-bin/vizier

[6] http://www.regulusastro.com/regulus/papers/m31/index.html

[7] Ciardullo, Robin, et al. “The Nova Rate in the Elliptical Component of NGC5128.” The Astronomical Journal.  April 1990.

[8] “The Andromeda Stream: A Giant Trail of Tidal Stellar Debris in the Halo of M31”. M. Irwin et al. 5 October 2001. http://www.ing.iac.es/PR/newsletter/news5/science1.html

[9] AAVSO-HOA

Bibliography    Data Table    Images