Lecture 2

The Hertzsprung-Russell Diagram

Blackbody Radiation
and Stellar Mass Determination

Glatzmaier and Krumholz 2

Prialnik 1.4
Pols 1
Andromeda and
Milky Way
collide in 4
billion years.
Approaching us
at 300 km/s
(Doppler shift)

HST astrometry
plus Doppler
(plus computer)

http://www.nasa.gov/mission_pages/hubble/science/milky-way-collide.html
 Back to the stars

In addition to location, brightness, and time variability

a key way we learn about stars is by analyzing
their light. At the most basic level we can analyze
their color, which turns out to be determined by
their broad band emission across wavelength
(black body emission)

In more detail we learn (a lot) more from analyzing

their spectra (transitions in individual atoms).
 E.g., ORION

" Even with the unaided eye

you should be able to
discern a difference
in color between
Betelgeuse and Rigel

Betelgeuse and Rigel are  !

" and -Orionis
Astronomers historically have measured the color of
a star by the difference in its brightness (magnitude)
in two images, one with a blue filter (B) and another
with a visual filter (V). (i.e., B = mB; V = mV)

This difference, denoted (B-V), is a crude measure of

the temperature.

Note that the bluer the object, the smaller

B will be (small magnitudes mean greater fluxes),
so small or more negative (B-V) means bluer and
hence hotter temperature.
 Blue FREQUENTLY USED FILTERS
filter ON THE TELESCOPE
(there are many more)

o
1 A = 1 Angstrom = 10 −8 cm

Today there are many more filters, including especially infrared,

but these are representative
For local stars – a “volume limited sample”

More luminous

Less
luminous

bluer redder
What does this mean?

• Most, though not all “stars” lie on a well defined

“line”
• The line is not straight
• There are many more points to the right of the
sun than to the left

To better understand this result, we need to know

what color means and what sets the color of a star
 As we will discuss more later, stars are blackbodies. This
is because they are extremely optically thick and the
light is trapped within for a long time, eventually coming
into equilibrium with its surroundings. Then the distribution
of energy with wavelength or frequency is given by the Planck
function:

2hc 2
dλ c
Bλ (T )d λ = 5 hc /( λk T ) d λ = − 2 dν
λ e B
−1 ν
2hν dν
3
Bν (T )dν = 2 hν /(k T )
c e B −1

This distribution gives the power emitted from the surface,

per unit “everything” – i.e., projected area of emitting surface,
per unit solid angle, per unit (frequency,wavelength)
Aside

We will see this Planck function several times later

so it is worth taking a look at it. It has 4 parts
1) The distribution function, aka the Bose Einstein
−1
⎛ kT
hν
⎞
distribution function ⎜ e − 1⎟ . It is common
⎝ ⎠
to see exponentials like this in ditribution functions.
They say how particles with energy (hν here)distribute
themselves among states whose characteristic scale is
is kT. The -1 is peculiar to spin 0 particles that can
have any number in a given state - subject to energy
conservation.

2) This number distribution function is multiplied by hν , the

energy of a particle
 3) And then by 4π (ν /c)2 d(ν / c) = 4π k 2dk, where k, the wave
number (1/λ ) is the "phase space" for paking so many photons
into a given volume

4) And by 2 because there are two polarization states for

the photon

ν 2dν ⎛ 1 ⎞
c ⎝e
( )
This gives (2)(4π 3 ) hν ⎜ hν /kT
− 1⎟⎠
8π hν 3dν 8π hν 3
or uν dν = and uν = erg cm−3 Hz −1
( )
c 3 e hν /kT − 1 ( )
c 3 e hν /kT − 1
c 2hν 3
Bν = uν = 2 hν /kT erg cm−2 s−1 Hz −1 Ster −1
4π c e (
−1 )
http://disciplinas.stoa.usp.br/pluginfile.php/48089/course/section/16461/qsp_chapter10-plank.pdf
https://www.youtube.com/watch?v=syQbWP-7WC4
 Blackbody (Thermal) Radiation

Intensity
As T rises:

• more radiation at
all wavelengths

• shift of peak emission

to shorter wavelength
quantum
• greater total emission cut-off
(area under the curve)
classic
A SMALLER COLOR INDEX MEANS A HOTTER STAR

(B −V )

= 0.65
Can also get the bolometric correction from
the Planck function
 Color index can be converted to temperature and absolute magnitude
to luminosity to give the Hertzsprung-Russel diagram in more physical units
Hertzsprung–Russell
diagram with 22,000
nearby stars from the
Hipparcos catalog
supplemented with
1000 stars from
other catalogs.

Absolute bolometric
magnitude has been
converted into
luminosity,

Note that the stars

prefer to congregate
in well defined strips
 What does it mean?

As we will discuss further in lectures 4 – 6, stars are

very optically thick and their radiation is trapped
inside them for a long time. The emergent light, to
good accuracy thus follows the Planck function
which describes radiation that is in equilibrium with
its surroundings:
2hν 3 dν
Bν (T )dν = 2 hν /(k T ) erg cm−2 s−1 Hz −1 steradian
c e B −1

1) Integrating over solid angle and frequencies

(a double integral) will give the total power
radiated per square cm by a blackbody

2) The frequency, or wavelength where most of the

power comes out can be obtained by setting the
derivative to zero.
 θ

To observer

Bν , the Planck radiance, is isotropic, but one

must correct for the orientation of different regions
of the stellar surface with respect to the observer.
The emission from each cm2 should be multipled by
Cos θ
 n

θ dΩ

Perhaps it s easier to visualize photons falling from the sky

from all directions on a flat area, dA,
on the surface of the Earth
Solid angle as a function of  and "

2π π /2

2∫ ∫ dΩ = 4π
0 0
 The total power emitted at all angles and frequencies
is thus:

P = ∫ dν ∫ dΩ Bν Cos θ dΩ = sinθ dθ dφ d Cosθ = − Sinθ dθ

0 hemi
sphere
∞ π /2 2π ∞ π /2 2π

P = ∫ dν ∫ dθ ∫ dφ B ν
Cos θ Sin θ = ∫ Bν dν ∫ dθ ∫ dφ Cos θ Sin θ
0 0 0 0 0 0
∞ ∞
0
⎛ 1⎞
= − 2π ∫ Bν dν ∫ Cos θ (dCos θ ) = 2π ⎜ 0 − ⎟ ∫ Bν dν
0 1 ⎝ 2⎠ 0
∞

=π ∫B
0
ν
dν
∞ ∞ 3 ∞ 3
2hν 3
dν k T ⎛ hν ⎞
3
dν
∫0 Bν dν = ∫0 c 2 hν
= 2 2 2 ∫⎜
h c 0 ⎝ kT ⎟⎠ hν
e −1 kT
−1
e kT

4 4 ∞
kT dx hν
= 2 3 2 ∫x x
3
x=
h c 0 e −1 kT
k 4T 4 ⎛ π 4 ⎞
= 2 3 2⎜ ⎟
h c ⎝ 15 ⎠
Adding the additional factor of π from the integral over angles

k4 ⎛ π 5 ⎞ 4
P = 2 3 2 ⎜ ⎟ T ≡ σT 4
h c ⎝ 15 ⎠
 dBλ
The maximum occurs where = 0, which is given by
dλ
hc 1
λmax = where x is the solution of
x kT
xe x
− 5 = 0 or x = 4.96511423...
e −1
x

0.28978 cm
λmax =
T
2.8978 × 10 7 A
=
T
slope = 0

This is known as Wien’s Law B

and it relates the “color” of a star
to its temperature P()

max" "
 Some blackbody examples:

1) A spherical object with no internal heat source but shining

because of stored up heat
d
dt
( )
(heat capacity)*mass *T = − 4π r 2σ T 4

e.g., a white dwarf

2) A main sequence star in thermal equilibrium

M

∫ nuc
ε = π σ
2 4
dm 4 r T
0

3) Two parallel planes emitting black body radiation at

each other
T1>T2
Both planes emit and absorb
radiation perfectly. T1 and T2
T1 T2 evolve depending on the
situation (sizes, masses, etc)
Some blackbody examples:

4) Planets

Planet absorbs some small

fraction of the power emitted
by a star, reprocesses it into
heat and radiates as a black
body with a temperature quite
different from that of the star.

The interior of the planet does not participate on relevant

time scales but the surface and atmosphere quickly
acquire a temperature as needed to satisfy balanced
lower – energy absorbed = energy radiated.
Since stars are good blackbodies and are to good
approximation, and are spheres, their luminosities are to
good accuracy

L = 4π Rstar
2
σ Tphoto
4

This provides both a tool for measuring stellar

radii (if we can measure the temperature and luminosity)
and a physical basis for understanding the systematics
of plots of luminosity vs temperature, i.e., the
Herzsprung-Russell Diagram
The sun vs a 5790 K blackbody
L = 4π Rstar
2
σ Tphoto
4

e.g. If L is high and T is small, R must be big

low high small
etc
 On the main sequence
more massive stars
have bigger radii.

It turns out that the

radius does not increase
quite as rapidly
as the mass. The total
range of variation
is about 100.

But there are other

kinds of stars with
bigger and smaller
radii.
Getting Masses in
Binary Systems
 Binary and Multiple Stars
(about one-hird to one-half of all stars)

Beta-Cygnus (also known as Alberio)

Alpha Ursa Minoris (Polaris)
Separation 34.6 . Magnitudes 3.0 and 5.3.
Separation 18.3 . Magnitudes
Yellow and blue. 380 ly away.
2.0 and 9.0. Now known to be a triple.
P > 75000 y. The brighter yellow component
Separation ~2000 AU for distant pair.
is also a (close) binary. P ~ 100 yr.
 Polaris

1.2 Msun Polaris Ab

Type F6 - V
4.5 Msun Polaris A
Cepheid

Period 30 yr

Polaris B is
F3 - V
 > 105 years
585 years 1165 years

Epsilon Lyra – a double double.

The stars on the left are separated by 2.3 about 140 AU;
those on the right by 2.6 . The two pairs are separated
by about 208 (13,000 AU separation, 0.16 ly between
the two pairs, all about 162 ly distant). Each pair would
be about as bright as the quarter moon viewed from the other.
 CLASSES OF BINARIES

• Visual binaries in which the stars and their orbital

elements are well resolved (in principle if one
waits long enough)

• Spectroscopic binaries – in which the presence of

a companion can be inferred by the periodic
Doppler shift exhibited in one or both spectra

• Eclipsing spectroscopic binaries in which the

spectrum shows evidence for binarity and the
light curve shows periodic eclipses or partial
occultations.
 KEPLER’S LAWS

• Solution to the central force problem for

a 1/r2 force, i.e., gravity
GM
r = - 3
r
r
• Orbits are ellipses with central force at one
focus of the ellipse

• A lines connecting the central force to the orbiting

body sweeps out equal areas in equal times

• The square of the period is proportional to the

cube of the semi-major axis. P 2 = 4π a3
2

GM

• Can be generalized to binary stars

Circular Orbit – Unequal masses
Two stars of similar mass
but eccentric orbits
Two stars of unequal mass and
an eccentric orbit
E.g. A binary consisting of
a F0v and M0v star

http://www.astronomy.ohio-state.edu/~pogge/Ast162/Movies/ - visbin
Some things to note:

• The system has only one period. The time for star A
to go round B is the same as for B to go round A.

• A line connecting the centers of A and B always

passes through the center of mass of the system.

• The orbits of the two stars are similar ellipses with

the center of mass at a focal point for both ellipses.

• At each point in time, the product of the mass of one

star times its distance to the center of mass is equal
to a similar product for the other star.
However:

The actual separation between the stars is obviously

not constant in the general case.

While not obvious, the separation at closest approach

is the sum of the semi-major axes of the two elliptical orbits,
a = a1+a2, times (1-e) where e is the eccentricity. [the
eccentricity of an ellipse is one half the distance between
the two focii divided by the semimajor axis)

At the most distant point the separation is a times (1+e).

For circular orbits e = 0 and the separation is constant.

 ASSUME FOR NOW CIRCULAR ORBITS

x
r1 r2 m2
CM
m1
both stars feel
the same gravitational m1v12 m2 v2 2 2π r1 2π r2
attraction and thus = = = Period
both have the same r1 r2 v1 v2
centrifugal force
Gm1m2 v r
= ∴ 1 = 1
(r1 + r2 )2 v2 r2
m1r12 v22 m2 v22
2
=
r1r2 r2 r1 m2 v1
More massive star is
closer to the center m1r1 = m2 r2 = =
r2 m1 v2
of mass and moves
slower. Also since the periods are
equal m1v1 = m2v 2
Circular Orbit – Unequal masses
KEPLER S THIRD LAW FOR BINARIES

GM1M 2 M1v12
=
( r1 +r2 )2 r1
r1 M1
M2 x
GM1M 2 M 2 v22 r2
=
+ (r1 +r2 ) 2
r2

G(M1 +M 2 ) v12 v22 4π 2 r12 4π 2 r22

2
= + = 2 + 2
(r1 +r2 ) r1 r2 P r1 P r2
4π 2
= 2 (r1 +r2 )
P
2 3 4π 2
P = K (r1 +r2 ) K=
G(M1 +M 2 )
 4π 2
( M1 + M 2 ) = 2
(r1 + r2 ) 3

GP

4π 2
M = 2
( AU ) 3 for the earth
G(1 yr)

Divide the two equations

 ⎛ r +r
( ) ⎞
3
M1 + M 2
=⎜ AU ⎟
1 2

M ⎜ 2
Pyr ⎟
⎝ ⎠
M1 r2 M 1 v2
= or =
M2 r1 M 2 v1

If you know r1, r2 in AU, or v1, v2, and P

in years you can solve for the two masses.
 The general case
(unequal masses eccentric orbits)
Define a coordinate system based on the center of mass
 
M1r1 + M 2r2 = 0
 M 
Then r1 = − 2 r2
M1
 
where r1 and r2 are the distances from the center of mass to
  
stars 1 and 2 respectively. Let r = r2 − r1 be the vector between
the two stars (in magnitude r = r1 + r2 since they are always in
opposite directions).

 ⎛ M2 ⎞  ⎛ M + M2 ⎞ 
r = ⎜ 1+ ⎟ r2 =⎜ 1 ⎟ r2
⎝ M1 ⎠ ⎝ M1 ⎠
 
 ⎛ M1 ⎞  ⎛ M 2M1 ⎞ r µr
r2 = ⎜ ⎟ r =⎜ ⎟ =
⎝ M1 + M 2 ⎠ ⎝ M1 + M 2 ⎠ M 2 M 2
where µ is the "reduced mass" (smaller than M1 or M2 ).

 µr
SImilarly r1 = − in the frame with the c/m at the origin.
M1
 M1
= 0.5, e = 0.2
M2

M1

r1

r 2 M2
“a” = average of
greatest and least
separations
 GM1M 2
M 2r2 = − r (1) v 22
r3 In the circular case r2 =
r2
GM1M 2
M1r1 = r (2) r = r1 + r2
3
r

Multiply (1) by M1 and (2) by M2 and subtract (2) from (1)

GM1M 2 (M1 + M 2 )
(
M1M 2 r2 − r1 = − ) r 3
r

GM1M 2 (M1 + M 2 )
M1M 2r = − 3
r
r
G(M1 + M 2 )
r = − 3
r
r
 Thus we have transformed to an equivalent central force
problem in which the mass that appears is the sum of the
masses and the relevant vector is the distance between the
two stars

 GM M 
f= 1 2
3
r
r
2
d r G(M1 + M 2 ) 
2
=− 3
r
dt r
  
r = r2 − r1
M1M2
µ=
M1 + M2
 µ   µ 
r1 = − r r2 = r
M1 M2

http://farside.ph.utexas.edu/teaching/336k/newton/node50.html
 The solution to the central force problem, as before,
is that the orbit is still an ellipse with distance between the
two stars given by
 
r = r2 − r1 =
(
a 1− e 2 ) e = 0 is
the circular case
1+ eCos θ in which r = r1+r2
Since (1-e 2 ) =(1+ e)(1− e), the semi-major axis of this ellipse is
half the sum of the maximum and minimum separations (θ =0,π )
1 ⎛
⎜
(
a 1− e ) (
2

+
a 1− e 2
) ⎞⎟ = 1 (a(1− e) + a(1+ e)) = a
2 ⎜⎝ 1+ e 1− e ⎟⎠ 2
1
2
( )
rmin + rmax = a

The orbital period is related to a by

4π 2
P2 = a3 where M =(M1 + M 2 )
GM
as we found for circular orbits. Also still m1v1 = m2v 2
and m1r1 = m2r2 though v and r both vary with time now.
The quantities to be measured then are a) the period
b) some measure of the semimajor axis and c) an observable
ratio, r1/r2 or v1/v2, that can give the ratio of the
masses.
M2 dr1 M 2 dr2
r1 = r2 and so =
M1 dt M1 dt
M1 v1 = M 2 v 2

at all times (and at all angles)

In the ideal case we can look at the orbit face on and

measure the closest and most distant separation of
the stars at Cos =0,  as discussed on the previous page,
rmin + rmax = 2a

Otherwise we have to measure r or v indirectly and also

somehow account for inclination.
 Spectroscopic binaries

Hubble Space Telescope photo of Gliese 623,

two stars separated by 2 AU.
The general case can
be solved but can be
quite complicated.

For this class we will

restrict the examples
to circular orbits.

One can extract “a”

as well as (sometimes)
information on
inclination.
 SPECTROSCOPIC BINARY MASSES
v1
Assume circular orbit. Measure
• Period
• Velocity of each star
• v is constant for circular orbit m1
r1
. r2 m2

v2
First get r1 and r2 from v1 and v2
vP
ri = i
2π 2π r
P=
Example: v
v1 = 75 km s −1 v2 = 25 km s −1
P= 17.5 days
Assume v is measured in plane of orbit,
otherwise we just see the component of the
velocity directed towards or away from us
 P
a = r1 + r2 =
2π
(
v1 + v 2 )
⎛ 17days ⎞
=⎜
⎝ 2π ⎠ ⎟( 100 kms −1
= )
2.34 × 1012
cm

= 0.156 AU

P = 17 d = 0.0465 years
−1
⎛ M1 + M 2 ⎞ ⎛ M1 + M 2 ⎞ (0.156)3
(0.0156)
3
2
P (yr )= ⎜ ⎟ ⇒⎜ ⎟ =
( )
2
⎝ M  ⎠ ⎝ M  ⎠ 0.0465
=1.76
Ratio of masses is v1 / v 2 = 3
4x =1.76 ⇒ M1 = 0.44M M 2 = 1.32M
The larger mass has the slower speed.
 Observer

v
•

θ = 0 face on, measure no velocity

π
θ = edge on, measure full velocity
2
 Eclipsing Binaries
(usually have circular orbits)

For an eclipsing binary you know you are viewing

the system in the plane of the orbit. I.e., Sin i = 1
Two views of the mass-luminosity relation (Malkov 2007,
left, and Henry 2004, right)
Recall
f Mq
τ MS ≈ n ≈ 3 − 4 for lower main sequence stars
Mn
2
1−n
⎛ M ⎞
∝M  1010
yr ⎜ ⎟
⎝ M ⎠
 HR diagrams for open clusters
M 67 and NGC 188

Schematic representation of
HR diagrams and main
sequence turn-offs observed
for different open clusters
horizonal
branch
 µ µ ⎛ a(1− e 2 ) ⎞ µ µ ⎛ a(1− e 2 ) ⎞
r1 = r= r2 = r=
M1 M1 ⎜⎝ 1+ eCosθ ⎟⎠ M2 M 2 ⎜⎝ 1+ eCosθ ⎟⎠
1 µ 1 µ
a1 =
2
( )
r1(0) + r1(π ) =
M1
a a2 =
2
( )
r2 (0) + r2 (π ) =
M2
a

a1 + a2 = a