Low Mass X-Ray Binaries and Cataclysmic Variables

Abstract

I'll review a few aspects of a class of binary stars in which the secondary- the donor, is a low mass star which is still on the main sequence or about to drift away from the main sequence. The primary is a highly evolved star, mostly a white dwarf (CV's) or a Neutron Star (LMXB's). The secondary fills up the Roche lobe resulting in mass transfer to the compact primary thru a stream and accretion disk.
The formation of binary stars in general is an open subject and not very well understood. So we'd consider only the evolution of CV's once they are somehow formed. We shall see that the evolution is primarily dictated by angular momentum loss mechanisms and the evolution of the secondary star, as it tries to adjust its radius to the mass loss (due to mass transfer and/or winds). Thus the relevant time-scales are angular momentum loss time-scales , Kevin-Helmholtz time-scales and also mass loss time-scales.
Though potentially relevant, the details of Accretion disk formation and/or transport of angular momentum in it wont be considered.

1  Introduction

Cataclysmic Variables (CV's) are a class of short-period binary systems which have a low mass secondary transferring mass onto a White Dwarf (WD) primary via Roche lobe overflow. The evolution of the CV is governed by angular momentum losses, and the dynamical responses of the binary system and the secondary to the mass transfer. The angular momentum loss is dominated by magnetic braking (at longer periods) and gravitational radiation (at shorter periods). Though the periods of CV's extend from approx. 1.3 hrs to 10 hrs, empirically it is seen that there is a distinct dearth of such systems between 2 and 3 hrs. One of the aims of any theory about CV's is to explain this 'period gap'. Another question that needs to be addressed is whether systems evolves across this period gap or the two populations above and below the gap are of different origin.
In general, a given CV will evolve from a higher mass secondary with a long period to a lower mass secondary, short period system. Presently, it is believed that they evolve through the period gap, becoming 'invisible' in the gap, and re-appearing at the other end of the gap. This is achieved by modelling the secondary star is such a way as to be consistent with the above requirements. From time to time however, it has been speculated that the CV's on either side of the period gap may be entirely different populations.
In the following section, i'll outline the basic physics involved in the problem by setting up the equations for mass transfer and angular momentum loss. Then i intend to work out in detail the equations in case of non-conservative mass transfer, i.e. the effect of winds and inefficient mass transfer. I then intend to work on the possibility of a 'Circumbinary Disk' and its influence on the binary. Finally, i'll try and outline what work can be done in this field which would be interesting, relevant and manageable!

2  Basic Physics

We start with the orbital angular momentum equation from Kepler's law:
J = M1 M2 æ
è
G a

M
ö
ø
1/2
 
(1)
To study the time evolution, we log differentiate
×
J

J
=
×
M1

M1
+
×
M2

M2
+ 1

2
×
a

a
- 1

2
×
M

M
(2)
Here,
M1 is the Mass of the primary(accretor),
M2 is the Mass of the secondary(donor),
M = M1 + M2 is the total Mass of the system and,
a is the binary separation.
(2) gives the functional dependance of the rate at which angular momentum is lost from the system in terms of the Masses involved and the separation. However, one can write the angular momentum loss rate in terms of the various 'modes' in which it can be lost as :
d J

d t
= æ
è
J

t
ö
ø

GR 
+ æ
è
J

t
ö
ø

[M\dot] 
(3)
The first term GR, represents the angular momentum loss rate as a result of gravitational radiation1. [M\dot] represents the total mass loss rate from the system. This may be in the form of an isotropic wind from the secondary (w2), or wind from a 'circumstellar' disk or ring as matter accretes onto the primary (w1), or may also be due to mass loss from a 'circumbinary' ring (cb) [1]. Thus, i can write :
×
M
 
=
×
M
 

w1 
+
×
M
 

w2 
+
×
Mcb
 
(4)
and consequently2,
æ
è
J

t
ö
ø

[M\dot] 
= æ
è
J

t
ö
ø

w1 
+ æ
è
J

t
ö
ø

w2 
+ æ
è
J

t
ö
ø

cb 
(5)
These considerations are important because if the mass loss in any of these modes is significant, the overall evolution of the system will be correspondingly affected - the specific angular momentum lost by the system depends on which of these modes dominates.
Each of these modes can be parameterized by associating a dimensionless specific angular momentum parameter with it. For example, in the case of 'Conservative' mass transfer, all these parameters are set to zero. Let us specialize in the 'Conservative' mass transfer case, and later we can generalize to the non-conservative case.

2.1  Conservative Mass Transfer Evolution

The equations governing the evolution of CV's is simplified greatly in this case, since we can set :

×
M
 
= 0,
×
M
 

1 
= -
×
M
 

2 
(6)
Consequently, (2) implies that the separation of the binary shrinks as the angular momentum is lost via Gravitational radiation, since the second term in (3) is now 0. The rate at which angular momentum is lost by this mechanism is calculated  [2], and is given in the weak field limit by :
æ
è
×
J

J
ö
ø

GR 
= - 32

5
G3

c5
M1 M2 M

a4
(7)
To estimate in what regimes this mechanism is most efficient, we recall the emperical/numerical relation :

RL

a
= 2

34/3
æ
è
q

1+q
ö
ø
1/3
 
= 0.462 æ
è
M2

M
ö
ø
1/3
 
(8)
Here,
RL is the Roche-lobe Radius,
q = M2/M1 is the mass ratio.
Using (8) to eliminate a in (7) and stable mass transfer, i.e. RL = R2, we get

æ
è
×
J

J
ö
ø

GR 
µ M1

M1/3
M27/3

R24
(9)
Now, if the secondary is near the main sequence, R2 µ M2, and so, ([J\dot]/J)GR ~ M2-5/3 µ P-5/3. Also, log differentiating (8), using (2) , we get :
×
R
 

L 

RL
=
2
×
J
 

J
+
2(-
×
M
 

2 
)

M2
æ
è
5

6
- M2

M1
ö
ø
(10)
Again, assuming R2 µ M2 and RL = R2, we get [3] :
-
×
M
 

2 

M2
=
-
×
J
 
/J

4/3 - M2/M1
(11)
Now, using (9) and (11), we obtain...
-
×
M
 

2 
=
×
M
 

GR 
@ 10-10 æ
è
Ph

2
ö
ø
-2/3
 
M\odot yr-1
(12)
Equation (12) gives the estimate of mass transfer rates driven by Gravitational Radiation. This mass transfer rate, in turn, dictates how luminous the system is. The Accretion Luminosity is given by  [4] :
Lacc =
G M1(-
×
M
 

2 
)

R1
(13)
Though a direct measurement of -[M\dot]2 is not possible to fix Lacc, (12) implies that for periods ~ 2 hrs., -[M\dot]2 is consistent with expected Lacc values( ~ 1032 erg/s) for M1 ~ 1 M\odot, R1 ~ 109 cm for CV's.
However, the mass transfer rates for periods > 2 hrs, arent sufficient enough to account for the expected luminosities. Moreover, 12 implies that higher period binaries would have lower mass transfer rates and thus lower luminosity. This is contradictory to what is observed. Longer period binaries are significantly more luminous than shorter period ones.
Thus we can conclude that at shorter periods ( £ 2hrs) angular momentum losses by Gravitational radiation are sufficient to drive mass transfer rates which can account for the expected luminosities. For longer periods though, we need some other mechanism for angular momentum loss, which can drive the mass transfer rates to higher levels, and in turn the luminosity. That mechanism is magnetic stellar wind braking, or simply magnetic braking.

2.2  Stellar Winds and Non-Conservative Mass Transfer

We've seen that angular momentum loss due to gravitational radiation is not sufficient to explain the behavior of CV's above P ³ 2 hrs. In fact, as mentioned earlier we dont see many CV's in the 2-3 hr range. Above 3 hrs, the mass transfer rates have to be much more than what GR can account for to explain the expected luminosities. So we turn to the 2nd mode for angular momentum loss in (3), mass loss3, to dictate the evolution of the CV4.
Now, in the absence of a magnetic field we can ignore the spin (see below) on the stars and concentrate on the orbital angular momentum of the stars. Also lets set the mass lost due to winds by the primary, Mw1, to 0. Thus, we have a CV transferring mass from the secondary to the primary, and the secondary is also loosing mass via a stellar wind. The angular momentum lost due to this wind is characteristic of the material associated with the secondary. Also since the mass transfer is no longer conservative, the assumptions (6) no longer hold. Instead, we have ...
×
M
 

1t 
= -b
×
M
 

2t 
,
(14)
×
M
 
=
×
M
 

2w 
+
×
M
 

1t 
+
×
M
 

2t 
=
×
M
 

2w 
+ (1-b)
×
M
 

2t 
The net angular momentum loss from the system in such a situation can be given as ...
d J

d t
= æ
è
J

t
ö
ø

GR 
+ æ
è
J

t
ö
ø

[M\dot] 
+ æ
è
J

t
ö
ø

[M\dot]t 
(15)
Here, the last term now represents the angular momentum loss due to mass transfer in the presence of a magnetic field. The twisting of the field lines due to the rotating fluid in the disc helps generate a torque on the secondary5.
Explicitly substituting in the various term and using equations (14) into (15), we get
æ
è
×
J

J
ö
ø
= 1

J
é
ë
æ
è
J

t
ö
ø

GR 
+ aw
×
M
 

2w 
Wa22 +aL(1-b)
×
M
 

2t 
Wa12 + aD
×
M
 

2t 
W(a-RL)2 ù
û
(16)
Here,
the a's represent parameters that may depend on detailed modelling of that particular loss mechanism, and the a's represent the length scale or lever arms for that mechanism. The subscripts w, L and D stand for 'wind', 'Loss' and 'Direct' respectively. Also,
J = m
Ö
 

G M a
 
,
m = M1 M2

M
= M2 æ
è
1

1+q
ö
ø
.
Substituting these equations into (16) we get ...
æ
è
×
J

J
ö
ø
= æ
è
×
J

J
ö
ø

GR 
+
aw
×
M
 

2w 

M2
(1+q) æ
è
a2

a1
ö
ø
2
 
+
aL(1-b)
×
M
 

2t 

M2
(1+q)+
aD
×
M
 

2t 

M2
(1+q) æ
è
a-RL

a
ö
ø
2
 
(17)

Now, the log derivative of (8) is
×
R
 

L 

RL
=
×
a

a
- 1

3
×
M

M
+ 1

3
×
M
 

2 

M2
(18)

Solving (2) for [a\dot]/a, and substituting from (17) we get:
×
a

a
= 2 æ
è
×
J

J
ö
ø

GR 
+ 2
×
M
 

2w 

M2
é
ë
aw (1+q) æ
è
a2

a1
ö
ø
2
 
- 1 + 1

2
q

1+q
ù
û
+
2
×
M
 

2t 

M2
é
ë
aL(1-b)(1+q)+ q b- 1+ 1

2
q

1+q
(1-b) + aD (1+q) æ
è
a-RL

a
ö
ø
2
 
ù
û
(19)

Now substituting (19) into (18)
×
R
 

L 

RL
= 2 æ
è
×
J

J
ö
ø

GR 
+
×
M
 

2w 

M2
é
ë
aw (1+q) æ
è
a2

a1
ö
ø
2
 
- 1 + 1

2
q

1+q
+ 1

6(1+q)
ù
û
+ 2
×
M
 

2t 

M2
é
ë
aL(1-b)(1+q)+ q b- 1 + 1

2
q

1+q
(1-b) + aD (1+q) æ
è
a-RL

a
ö
ø
2
 
+ b

6
q

1+q
+ 1

6
1

1+q
ù
û
(20)
Now, the secondary is a main sequence star and so we can exploit the mass-radius relation for MS stars along with the contact condition which we assume holds for our binary system, so that ...
×
R
 

2 

R2
= xMS
×
M
 

2 

M2
Þ
×
R
 

L 

RL
= xMS
×
M
 

2 

M2
= xMS é
ë
×
M
 

2w 

M2
+
×
M
 

2t 

M2
ù
û
(21)

Now, substitute (21) on the LHS of (20) and rearrange to get:
×
M
 

2t 

M2
=
2 æ
è
×
J

J
ö
ø

GR 
-
×
M
 

2w 

M2
é
ë
xMS - {aw (1+q) æ
è
a2

a1
ö
ø
2
 
- 1 + q

2(1+q)
+ 1

6(1+q)
} ù
û

2 é
ë
xMS-{aL(1-b)(1+q)+ q b- 1 + q(1-b)

2(1+q)
+ aD (1+q) æ
è
a-RL

a
ö
ø
2
 
+ bq +1

6(1+q)
} ù
û
(22)

This then, is a most general description of the mass transfer rate as dictated by various angular momentum loss mechanisms, discounting the possible affects of a 'Circumbinary disk' which will be considered next.
· Reality Check:
To get to (11), which is the simplified, conservative counterpart of (22), we set:
[M\dot]2w = 0, xMS = 1, b = 1, aw = 1, aD = 06.
On rearranging, we indeed get back (11)!

3  The Circumbinary Disk

The circumbinary disk, can in principle extract infinite amount of angular momentum from the system [5]. The key is to able to cut off this mechanism, along with the other mechanisms, at the upper end of the period gap. This prolly implies magnetic coupling between the disk and the binary. Also, the disk has to be maintained for it to be relevant for long term evolution. Thus the disk may be static(extending outward to infinity?) or may be fed by mass flowing out of the binary [6]. On the other hand it should'nt be too bright, so as to be observable easily (since we dont seem to see any such disk's around binaries too often!). The angular momentum transport within this disk thus also needs to be investigated [7].
I am working on all this....

4  Nova-like events

Initially thought i'd look at what people have done here, but the circumbinary disk is more interesting. Important thing to remember here is the very short time scales at which a lot of mass can be lost from the system and so also angular momentum. The system may react violently, even be disrupted if the secondary is burping from time to time [8]!

5  Conclusions

I need to generalize (22) further to include the circumbinary disk. In spite of not doing so (and its apparent ugliness!), (22) is the most general form of the mass transfer rate in terms of various angular momentum loss mechanisms. Detailed modelling would be required to set the various parameters, like the a's and the lever arms (the a's).

References

[1]
Soberman et al, astro-ph/9703016.
[2]
Landau & Lifschitz.
[3]
King,A.R., Q. Jl R. astr. Soc. (1988), 29, 1-25
[4]
Frank, King, Raine.
[5]
Pringle, P.E.Mon. Not. R. astr. Soc. (1991), 248, 754-759.
[6]
Spruit & Taam, ApJ (2001), 548, 900-907.
[7]
Taam & Spruit, ApJ (2001), 561, 329-336.
[8]
Kato & Hachisu, ApJ (1991), 373, 620-623.

Footnotes:

1GR experts, please opine!
2Why Partials?
3i treat 'winds' as essentially a loss of mass from the system
4the circumbinary ring is too hot to handle! lets leave that for later.
5Why isnt this term present in the conservative case?
6Actually, this term should've been included in the Conservative case, but since i didnt, i'am setting it to zero.
  • Back Home