Sie sind auf Seite 1von 13

Gravo-thermal Catastrophe

Surya Prakash
IIT Kanpur, India

Abstract
This report presents analysis of small part of paper ”The Gravo-Thermal
Catastrophe in Isothermal Spheres” published by D. Lynden-Bell, Roger
Wood. In this report, we will get familiarized with concept of gravother-
mal catastrophe for any system. Then we will derive certain thermodynamic
properties of such system. Finally, based on the behaviour of such thermo-
dynamic variable we would justify certain properties of system undergoing
gravothermal catastrophe.
Keywords: Gravothermal Catastrophe, Isothermal, Equilibrium

1. Gravitating System and Isothermal Spherical System

S M 5 M3 M3
⇒ = − log(pβ 0 2 ) − log m + log(2π) + β 0 T (1)
k m m2 m2
For a system, purely under gravitational force, Virial theorem states that

2T + U = T + E = 0 ⇒ T = −E

Where E= total energy; T = kinetic energy; U = Potential energy For such


system increase in E causes a decrease in kinetic energy and therefore kinetic
energy per particle or temperature.As we know
dE
Cv =
dTs
Where Cv is specific heat and Ts is temperature of system. It is clear that
Cv is negative for such system.
Using Virial theorm to system with surface pressure

2T + U = 3pV

Preprint submitted to Journal Name November 8, 2019


where 0 p0 is surface pressure. For Isothermal system pV = constant and
therefore we can see as before that it has negative specific heat provided
system is sufficiently centrally condensed so that it can be regarded as held
by his own gravity in stead of surface pressure. This is condition is satisfied
by isothermal spheres with radius smaller than critical radius(will be defined
later)

2. Understanding Gravo-thermal Catastrophe


Let us understand this concept using an experiment. Suppose, a large
number ’N’ of point mass ’m’ are released under their mutual gravity inside
a sphere of radius ’re ’ having total energy E(< 0) and total mass ’M=Nm’.
We are defining critical radius(will be explained later) as
rc = 0.335GM 2 /(−E)
We start with our sphere in a box with re < rc and surround it with box
with radius greater than rc and suddenly remove the inner box. Release of
pressure results in expansion of system which cause adiabatic cooling. As
central part is held together by self-gravity do not expand as much as outer
part which are held by wall pressure. So central part cools less than outer part
and thus temperature gradient is set up. The inner part being self gravitating
have negative-specific heat.As they lose the heat they shrink and grow hotter
where as outer part have positive specific heat they receive heat and grow
hotter. If outer part has larger specific heat than the magnitude of specific
heat of inner part than conductive transfer of heat from central part will
raise the high central temperature faster than it raises the lower temperature
of outer part. This situation we call as Gravo-thermal catastrophe where no
equilibrium is possible.
2.1. How do we bring equilibrium?
As re was larger than critical radius we observed that centre seemed to
condense out and evolved towards high temperature and densities. Entropy
was continuously generated. In order to stop this we reduced the radius lower
than critical radius but, surprisingly, instead of returning to normal density, it
continued to get hotter. Pressure due to conduction of heat flux from central
part was accelerating very rapidly. With much exterior reinforcement, we
reduced the size to much smaller than critical radius only then we were able
to reverse the heat flux from outward to inward and system settled into a
high temperature equilibrium.

2
3. Mathematical derivation
Entropy of a system can be written as
XZ
S = −k f i logf i d6 τ (2)
i

Where
i)S is entropy
ii)k is Boltzmann’s constant
iii)The integration over phase space
iv) We have divided the mass points into group and f i is their number
density in phase space of ith group.f i is function of phase space and mi
is average mass of ith group.
v) Phase space is specified as position vector r= (x,y,z) and velocity vector
c = (u,v,w) therefore

d6 τ = dxdydz dudvdw

We are considering total angular momentum H = 0 for simplification of


calculation.
We define total phase space density at r, c to be f (r, c)= i mi f i (r, c).
P
The spatial density ρ(r) is then
Z
ρ(r) = f (r, c)d3 c

with integration over all velocities.


The kinetic energy is
XZ 1 Z
1 2
i 2 i 6
T = m c f (r, c)d τ = c f (r, c)d6 τ
i
2 2

The total potential energy

ρ(r)ρ(r0 ) 3 3 0 ff0 3 3 0
Z Z Z Z
G G
U =− d rd r = − d τd τ
2 |r − r0 | 2 |r − r0 |

Where f 0 = f (r0 , c0 )

3
Thus total energy is

ff0 3 3 0
Z Z Z
1 2 G
E =T +U = c f (r, c)d6 τ − d τd τ (3)
2 2 |r − r0 |

Total number of mass points in the ith mass group is


Z
N = f i (r, c)d6 τ
i
(4)

Now, we have to maximize ’S’ keeping energy ’E’ and numbers N i con-
stant. We will do this using Lagrange multipliers kβ 0 and kαi . Let’s define,
lagrangian 0 L0 as
" # " #
Z Z Z 0 Z
1 G f f X
L = S+kβ 0 E− c2 f (r, c)d6 τ + 0
d3 τ d3 τ 0 + kαi N i − f i (r, c)d6 τ
2 2 |r − r | i

"Z " # #
ff0
Z Z Z
1 2 G X
⇒ δL = δS−kβ 0 c δf d6 τ − δ 3 3 0
d τd τ − kα i
δf i d6 τ
2 2 |r − r0 | i

XZ
⇒ δL = −k δf i (logf i + 1)d6 τ
i
"Z Z Z " # #
0 0 Z
0 1 2 G f δf + δf f 3 0
X
−kβ c δf d6 τ − 3
d τd τ − kα i
δf i d6 τ
2 2 |r − r0 | i
(5)
0 0 0
As f, f are dummy indices therefore integration of f δf and δf f over
phase space is equivalent.So second part of above equation can be written as
"Z Z Z " # # "Z " # #
0 0 Z 0
1 2 G f δf + δf f 1 f
kβ 0 c δf d6 τ − 0
d3 τ d3 τ 0 = kβ 0 δf c2 −G 0
d3 τ 0 d3 τ
2 2 |r − r | 2 |r − r |

If we define gravitational potential caused by the density corresponding


to f as
ρ(r0 ) 3 0 f (r0 , c0 ) 3 0
Z Z
ψ(r) = G d r = G dτ (6)
|r − r0 | |r − r0 |

4
we can write above equation as
"Z " # # "Z " # #
Z 0
1 f X 1
kβ 0 δf c2 −G d3 τ 0 d3 τ = kβ 0 mi δf i c2 −ψ(r) d3 τ
2 |r − r0 | i
2
(7)
Thus equation (4) can be written as
" ! #
XZ 2
c
δL = −k δf i logf i + 1 + β 0 mi − ψ + α i d6 τ
i
2

In order to maximize ’S’ we will


!
c2
δL = 0 ⇒ logf i + 1 + β 0 mi −ψ + αi = 0
2
!
c2
⇒ −logf i = 1 + β 0 mi −ψ + αi
2
i 0 i 0 i
⇒ f i = e(−1−α ) e−β m  = Ai e−β  (8)
Where
" #
2
c c2
i = mi − ψ ; Ai = exp[−(1 + αi )];  = −ψ
2 2

Average kinetic energy of ’mi ’can be calculated as


 
0 i c2 2
exp − β m 2 mi c2 d3 c
R i i c2 3 R
f m 2d c 3
R
i 3
= R   = 0 (9)
fdc 2
exp − β 0 mi c d3 c 2β
2

Here, β 0 is identified as (kT )−1 .[ Explained in appendix I (page 13)] Integrat-


ing equation (7) we obtain the density of ith mass group

c2
Z Z
ρ = m f d c = mi Ai exp(β 0 mi ( − ψ)d3 c, r < re
i i i 3
2
Z
0 i c2
 2π  32
i i −β 0 mi ψ 0 i
⇒ ρi = A m e eβ m 2 d3 c = Ai mi e−β m ψ
mi β 0

5
0 i
⇒ ρi = B i e−β m ψ
where  2π  32
i i i
B =Am
mi β 0
The total density is
X X
ρ= ρi = B i exp(β 0 mi ψ), r < re
i i

So the potential is
0 i 0
ρ(r0 ) 3 0
Z Z P i
i B exp[β m ψ(r )] 3 0
ψ(r) = G 0
d r = G dr
r0 <re |r − r | r0 <re |r − r0 |

This integral eqution for ψ may be solved by differentiation which gives


X
52 ψ(r) = −4πGρ(r) = −4πG B i exp(β 0 mi ψ), r < re ; (10)
i

Boundary condition ψ is order of 1/r at infinity and ∂ψ/∂r and ψ are


continuous at r = re .
Considering only spherically symmetrical solution and simple case of one
mass group only we get
1 d  2 dψ 
r = −4πGBexp(βψ), r < re (11)
r dr dr
[As this is general solution of Poisson’s equation.]

3.1. Calculation of thermodynamic parameters for bounded isothermal sphere


Using Virial theorm to system with surface pressure

2T + U = 3pV

where ’p’ is surface pressure. p = f 1 31 mc2 d3 c;(This can be derived from


R

kinetic theory of gases.) V = 34 πre 3 is the volume U is total potential energy


T is total kinetic energy From equipartition theorem we have
3 M 3M
T = 0
=
2 mβ 2 β

6
As total energy is
E =T +U
So using virial theorem
3M
E = 3pV − T = 3pV −
2 β
Similarly
M
U = 3pV − 3
β
which vanishes for perfect gas in absence of gravitation. Surface pressure can
be related to edge density using kinetic theory of gas(equipartition theorem)
and equation (8) for particular mass group.
Z Z
11 2 3 2 3 1 3 ρe ρe
p= f mc d c = f d c = = (12)
at re 3 3 2β 0 at re mβ 0 β

For one mass group equation (7) takes the form


h  c2 
f 1 = Aexp − β −ψ (13)
2
and at r=re
GM
ψ= (14)
re
So using equation (11),(12),(13)

ρe
Z h  c2 GM i  2π  32  GM 
= Aexp − β − d3 c = A exp β (15)
m 2 re β re

Using equation (1) and (12)


Z hZ Z 2 Z
1 c
i
1 1 6 1 6 6 1 6
S = −k f log f d τ = −k f log Ad τ − f β d τ + β f ψd τ
2
(16)
Hence using equation (3)
Z
M
= N = f (r, c)d6 τ
m

7
and definition of kinetic energy equation (15) can be written as
Z
S M 0 0
= − log A + β T − β ρψd3 r (17)
k m
Using equation (11) and (14) we are eliminating ’A’ in term of 0 p0

pβ  2π  32  GM 
=A exp β
m β re
 2π  32
0 GM
 
0
⇒ pβ = A exp mβ
mβ 0 re
5 3
 GM 
⇒ A = pβ 0 2 m3/2 (2π)− 2 exp − mβ 0
re
Putting the value of ’A’ in equation (16) we get

GM 2
Z
S M 5 M3 M3
= − log(pβ 0 2 ) − log m + log(2π) + β 0 + β 0T − β 0 ρψd3 r
k m m2 m2 re
Using the definition of ψ {equation (5)}
!
S M 5 M3 M3 GM 2
⇒ = − log(pβ 0 2 ) − log m + log(2π) + β 0 + T + 2U
k m m2 m2 re

2
S M 0 52 0 GM
  3M M3
⇒ = − log(pβ ) + β + 2E − β 0 T − log m + log(2π)
k m re 2m m2
Kinetic energy of gas system can be written as T = 23 N kTs and N = M
m
.
Using this we can write

S M 5
 GM 2  3M  M3
⇒ = − log(pβ 0 2 ) + β 0 + 2E − 1 + log m + log(2π)
k m re 2m m2
Eliminating β 0 with β

mS 5
 GM 2  3 2 M3
⇒ = −M log(pβ 2 ) + β + 2E − M (1 − logm) + log(2π)
k re 2 3 m2
(18)
Now if we define
v1 = β(ψ − ψ(0))

8
1 1
r1 = [4πGβBexp[ψ(0)β]] 2 r = (4πGρ0 β) 2 r
then density can be written as

ρ = ρ0 ev1

At the edge of sphere we are introducing a variable


1
z = r1 |r=re = (4πGρ0 β) 2 re

Thus density contrast between edge and centre is


ρe
= exp(v1 (z)) (19)
ρ0
and gavitational potential energy can be written as
 dψ  re  dv1 (r1 ) 
2
GM = − r =− r1
dr re β dr1 z

GM β
⇒ = −zv10 (20)
re
From here v1 and v10 are understood as v1 (z) and v10 (z)
From equation (11), (18)
ρe
ρe = ρ0 ev1 ; p =
β

and using equation (18) and definition of z

1 z2
z = (4πGρ0 β) 2 re ⇒ ρ0 =
4πGβre2

Using equation (11) and (19)

z2 ev1 GM 2 z 2 ev1
p= . = . (21)
4πGβre2 β 4πre4 (−zv10 )

GM 2 z 2 ev1
⇒ 3pV = . (22)
re (−zv10 )2

9
Using equation (19) replacing β and equation for energy from Virial theorem
E = 3pV − T = 3pV − 32 M β

GM 2 h z 2 ev1 3 1 i
E= − (23)
re (−zv10 )2 2 (−zv10 )
GM 2
Replacing β from equation (19); p from (20) and re
from (22) and putting
it into entropy equation (17), we get
mS 1 1 2Ere
= logV − 2logz − log(−zv10 ) + (−zv10 ) − v1 + (−zv10 ) + constant
kM 2 2 GM 2
(24)

4. Physical Explanation

10
This is a graph of −Ere /GM 2 and −v1 [from equation(22)] where solid line
shows the locus of possible equilibrium states(stable or unstable) based on
maximal and minimal of entropy. We are also showing entropy with this
equilibrium plot. For a given energy, it is clear from Fig 2 that entropy is
maximal for equilibrium states upto point ’A’ as this is inflection point. This
indicates that states corresponding to entropy below this point ’A’ are in
stable states. If we increase radius above ’A’, we reach to a point such as ’B’
where system evolve spontaneously towards a state with higher entropy but
this time there is no local maximal entropy to where it can go so it keeps
on increasing and system becomes unstable. Radius corresponding to this
inflection point is defined as critical radius.

If we compress the the box below critical radius, then E,re and v1 change
simultaneously. So, other dimension is added to Fig 2 which is log re [fig
3].where surface shows the locus of possible equilibrium states. We are show-
ing the intersection of entropy surface with this equilibrium. In the exper-
iment system tends to go to higher and higher v1 along the line of higher
and higher entropy.The current entropy surface cut the equilibrium surface
at much lower value of re than current re . Therefore, we need to reduce
the radius of system so that equilibrium state can be consistent with present
entropy. At the particular re fig 3 can be visualized as fig 2. So, in general,
initially, system can at point of higher −v1 i.e. according to diagram at some

11
point such as ’C’ and system will eventually result into same situation as
in case of ’B’(in previous case) but if we compress the system such that it
comes to position such as ’D’. The spontaneous reaction of system will be to
reach the maximal entropy and thus it will settle to that equilibrium point.

5. Acknowledgement
This is completely based on paper The Gravo-Thermal Catastrophe in
Isothermal Spheres published by D. Lynden-Bell, Roger Wood.

12
6. Appendix I
From equation (20) and (22)

GM 2
E=− q(z)
re
GM 2
p= g(z)
4πre4
where
z 2 ev1
g(z) =
(−zv10 )
3
q(z) = − g(z)
2(−zv10 )
The heat dQ is given by
GM 2 GM 2 0
dQ = dE + pdV = [g(z) + q(z)]dre − q (z)dz
re2 re
3 0
dQ 3 dre 2
q (z)
⇒ 2 GM 2
= − dz (25)
3 re
[g(z) + q(z)] 2 re g(z) + q(z)
But 32 [g(z) + q 0 (z)] = [−zv10 ]−1 and from equation (19)
h GM i−1
β= (−zv10 )
re
hence equation (24) can be written
h1 Z 0
3 q (z) i 1
βdQ = M d log V − dz = M d[ log V − I] (26)
2 2 g+q 2

z
q 0 (z)
Z
1
I= 2 dz = 2log z + log(−zv10 ) + 2q(−zv10 ) − (zv10 ) + v1
3
(g + q) 2
On comparing with equation(25) with equation(23) we can say that
kβ 0 dQ = dS
and from thermodynamics entropy definition here β 0 can be equated to
(kT )−1 .

13

Das könnte Ihnen auch gefallen