AST1100 Lecture Notes

3 Extrasolar planets
1

Detecting extrasolar planets

Most models of star formation tell us that the formation of planets is a

common process. We expect most stars to have planets orbiting them. Why
then, has only a very few planets (less than ten by fall 2009) around other
stars been seen directly? There are two main reasons for this:
1. The planets orbit is often close to the star. If the star is far away from
us, the angular distance between the star and the planet is so small
that the telescope cannot separate the two objects.
2. The light from the star is much brighter than the starlight reflected
from the planet. It is very difficult to detect a faint signal close to a
very bright source.
How large is the angular distance on the sky between Earth and Sun seen
from our closest star, Proxima Centauri 4.22 light years away? Look at the
geometry in figure 1. The distance r is 4.22 light years, the distance SunEarth d is 150 106 km. Using the small angle formula from geometry (and
this is indeed a very small angle),
d = r
we find = 0.00021 (check!). In astrophysics we usually specify small angles in terms of arcminutes and arcseconds, denoted and . There are 60
arcminutes in one degree and 60 arcseconds in one arcminute. Thus the
angular distance between Sun and Earth as seen from Proxima Centauri is
0.77 . From the ground, the best resolution a normal telescope can reach is
0.4 under very good atmospheric conditions (actually using so-called adaptive optics better resolutions may be attained). This means that two objects
with a smaller angular distance on the sky cannot be separated by the telescope. So the green men on a planet orbiting our nearest star would just
1

r
Figure 1: The angular extension of a distant planet with diameter d.

be able to see the Earth with the best telescopes under very good atmospheric conditions (provided the atmosphere on this planet is similar to the
Earths)! The Hubble Space Telescope which is not limited by the atmosphere can reach a resolution of 0.1 . For the people on a planet orbiting a
star located 100 light years away from Earth, the angular distance between
the Earth and the Sun would be 0.03. From this planet, our green friends
would be unable to see the Earth using the Hubble Space Telescope! The
only confirmed directly detected planet is orbiting its mother star at a distance of 55 AU (The distance Sun-Pluto is about 40 AU). A huge advance
in optics and telescope technology is needed in the future in order to resolve
planets which are orbiting closer to their mother star.
Still, about 360 planets orbiting other stars have been detected (by fall
2009). The reason for this can be found in the previous lecture: In a starplanet system, the planet and the star are orbiting their common center of
mass. Thus, the star is moving in an elliptical orbit. If the velocity of the
star can be measured, then a regular variation of the stars velocity as it
orbits the center of mass should be detected. This is the way most of the
extrasolar planets have been discovered so far.
One way to measure the velocity of a star is by the Doppler effect, that
electromagnetic waves (light) from the star change their wavelength depending on whether the star is moving towards us or away from us. When the
star is approaching, we observe light with shorter wavelength, the light is
blueshifted. On the contrary, when the star is receding, the light is redshifted.
By measuring the displacement of spectral lines in the stellar spectra (more
details about this in a later lecture), we can measure velocities of stars by

the impressive precision of 1m/s, the walking speed of a human being. In

this way, even small variations in the stars velocity can be measured. Recall
the formula for change in wavelength due to the Doppler effect
0
vr
= ,
0
c
where is the observed wavelength and 0 is the wavelength seen from the
rest frame of the object emitting the wave. There is one drawback of this
method: only radial velocity can be measured. Tangential velocity, movements perpendicular to the line of sight, does not produce any Doppler effect.
The orbital plane of a planet (which is the same as the orbital plane of the
star) will have a random orientation. We will therefore only be able to measure one component of the stars velocity, the radial velocity.
In figure 2 we have plotted the situation. The angle i is called the inclination of the orbit. It is simply the angle between the line of sight and a
line perpendicular to the orbital plane (see figure 2). When the inclination
i = 90 , the plane of the orbit is aligned with the line of sight and the velocity measured from the Doppler effect is the full velocity. For an inclination
i = 0 , there is no radial component of the velocity and no Doppler effect
is seen. A regular variation in a stars radial velocity could be the sign of a
planet orbiting it.
We will in the following assume circular orbits (i.e. the eccentricity e = 0).
This will make calculations easier, the distance from the center of mass a is
always the same and more importantly, the velocity v is the same for all
points in the orbit. In figure 3 we show how the radial velocity changes
during the orbit of the star around the center of mass. If the inclination is
i = 90 , then the radial velocity vr equals the real velocity v in the points B
and D in the figure. For other inclinations, the radial velocity vr in points B
and D is given by
|vr | = v sin i.
(1)
This is found by simple geometry, it is the component of the velocity vector taken along the line of sight (do you see this?). NOTE: The velocity
v discussed here is the orbital velocity of the star, i.e. the velocity of the
star with respect to the center of mass. Normally the star/planet system,
i.e. the center of mass, has a (approximately) constant velocity with respect
3

N
line of sight

CM

vr

v
v
~
Figure 2: Inclination: The angle between the line of sight and the normal N
to the orbital plane is called the inclination i. The maximum radial velocity
of the star equals v sin i.

v = 0
v r = v
D

v = v
vr = 0

line if sight

CM

v = v
vr = 0

B
v = 0
vr = v

vr

v
B

A
A

v
D

Figure 3: The velocity curve of a star orbiting the common center of mass
with a planet. The points where the component of the velocity vector along
the line of sight is zero (A and C) as well as the points where the radial
component equals the full velocity (B and D) are indicated. In the figure, we
have assumed an inclination if 90 .
5

to the observer. This velocity vpec is called the peculiar velocity and must
be subtracted in order to obtain the velocity with respect to the center of
mass. Recall from the prevous lectures that the velocity of the star can be
decomposed into the velocity of the center of mass (peculiar velocity) and
the velocity of the star with respect to the center of mass (which is the one
we need).

Determining the mass of extrasolar planets

We know that Keplers third law connects the orbital period P , the semimajor
axis a (radius in the case of a circular orbit) and mass m of the planet/star
(do you remember how?). From observations of the radial velocity of a star
we can determine the orbital period of the star/planet system. Is there a
way to combine this with Keplers laws in order to obtain the mass of the
planet? The goal of this section is to solve this problem. We will deduce a
way to determine the mass of an extrasolar planet with as little information
as possible.
In the following we will use m , a , v for mass, radius of the orbit and
velocity of the star in its orbit around the center of mass. Similarly we will
use mp , ap and vp for the corresponding quantities regarding the planet. The
constant velocities may be expressed as,
v =

2a
P

vp =

2ap
.
P

(2)

Note again that this is velocity with respect to center of mass, any peculiar
velocity has been subtracted. In the lecture notes for lecture 1-2, section 5,
we found expressions for the position of the two bodies m1 and m2 taken in
the center of mass frame, ~r1CM and ~r2CM . Before reading on, look back at
these lecture notes now and make sure you remember how these expressions
were obtained!
Did you check those lecture notes? Ok, then we can continue. Take these
masses to be the star and the planet. Using these expressions, we obtain
(check!)
|~rCM |
mp
a
=
=
,
|~rpCM |
m
ap

where the expressions for the semimajor axes a1 and a2 from lecture 1-2 were
used. Using equation (2), we also have that
a
v
vr / sin i
vr
=
=
,
=
ap
vp
vpr / sin i
vpr
where equation (1) was used. NOTE: Here, the radial velocities vr and
vpr refer to the velocity at the point B in figure 3, the point for which the
radial velocity is maximal. We may use these two equation to eliminate the
unknown velocity of the planet
vpr = vr

m
.
mp

(3)

We will now return to Keplers third law,

4 2 a3
m + mp = 2 ,
P G
where we have used the exact expression for Keplers third law, derived in
problem 2 in lecture notes 1-2. From section 5 in those notes, we also had
that
a = a + ap ,
the semimajor axis a (of the orbit of the planet seen from the star or vice
versa) equals the sum of the semimajor axes of the orbits of the planet and
star about the center of mass. We can now express these in terms of velocities
(equation 2)
P
a=
(v + vp ).
2
Inserting this into Keplers third law, we have
m + mp =

P
(v + vp )3 .
2G

Normally we are only able to measure radial velocities, not the absolute
velocity. We thus use equation (1) as well as equation (3) to obtain
3
m
P vr
P (vr + vpr )3
1+
m + mp =
=
3
3
2G
mp
sin i
2G sin i

!3

Assuming that the star is much more massive than the planet (which is
normally the case, for instance mJupiter /mSun 103 ) we get
3
P vr
m3
m =
,
2G sin3 i m3p

which solved for the mass of the planet (which is the quantity we are looking
for) gives
2/3
m vr P 1/3
.
mp sin i =
(2G)1/3
Normally, the mass of the star is known from spectroscopic measurements.
The radial velocity of the star and the orbital period can both be inferred
from measurements of the Doppler effect. Thus, the expression mp sin i can
be calculated. Unfortunately, we normally do not know the inclination angle
i. Therefore, this approach for measuring the planets mass can only put a
lower limit on the mass. By setting i = 90 we find mmin
p . If the inclination
angle is smaller, then the mass is always greater than this lower limit by a
factor of 1/ sin i. In the next section however, we will discuss a case in which
we can actually know the inclination angle.

Measuring the radius and the density of

extrasolar planets

If the inclination is close to i 90 , the planet passes in front of the stellar

disc and an eclipse occurs: The disc of the planet obscures a part of the
the light from the star. Be looking at the light curve of the star, a dip will
occur with regular intervals corresponding to the orbital period. In figure 4
we show a typical light curve. When the disc of the planet enters the disc of
the star, the light curve starts falling. When the entire disc of the planet is
inside the disc of the star, the light received from the star is now constant
but lower than before the eclipse. When the disc of the planet starts to leave
the disc of the star, the light curve starts rising again. When such a light
curve is observed for a star where a planet has been detected with the radial
velocity method described above, we know that the inclination of the orbit
is close to i = 90 and the mass estimate above is now a reliable estimate of
the planets mass rather than a lower limit.

flux

F
E

B
C

t
3

Figure 4: The lower part of the figure shows a planet (small filled circle)
eclipsing a star (large open circle). The upper part shows a plot of the flux
variation with time at the different points during the eclipse. The moments
at which the eclipse starts t = t0 and ends t = t3 as well as the moments
when the full disc of the planet enters t = t1 and leaves t = t2 the star are
indicated.

In these cases, where the effect of the eclipse can be seen, the radius of
the planet may also be measured. If we know the time of first contact (time
t0 in figure 4), the time when the disc of the planet has fully entered the disc
of the star (time t1 ) as well as the velocity of the planet with respect to the
star, we can measure the radius of the planet. It took the disc with diameter
2Rp of the planet a time t1 t0 to fully enter the disc of the star. The planet
moves with a velocity v + vp with respect to the star (the velocity vp is only
the velocity with respect to the center of mass). Using simply that distance
equals velocity times interval, we have
2Rp = (v + vp )(t1 t0 )
As we have seen, we can obtain t1 and t0 from the light curve. We can also
obtain the velocity of the planet (the velocity of the star is measured directly
by the Doppler effect) by using equation (3),
vp = v

m
.
mp

Here the mass of the planet mp has been calculated since we know that the
inclination is i 90 . Thus, the radius Rp of the planet is easily obtained.
Combining the measured mass and radius of the planet we get an estimate
of the mean density
mp
.
p =
4/3Rp3
We can use this to determine whether the detected planet is terrestrial planet
with a solid surface like the inner planets in the solar system, or a gas planet
consisting mainly of gas and liquids like the outer planets in our solar system.
The terrestrial planets in our solar system have densities of order 4 5 times
the density of water whereas the gas planets have densities of order 0.7 1.7
times the density of water. If the detected planet is a terrestrial planet, it
could also have life.
Finally, note that also the radius R of the observed star can be obtain
by the same method using the time it takes for the planet to cross the disc
of the star,
2R = (v + vp )(t2 t0 ).
We have discussed two ways of discovering extrasolar planets,
by measuring radial velocity
10

 by measuring the light curve

In the following problems you will also encounter a third way,
by measuring tangential velocity
For very close stars, the tangential movement of the star due to its motion
in the orbit about the center of mass may be seen directly on the sky. The
velocity we measure in this manner is the projection of the total velocity onto
the plane perpendicular to the line of sight. There are two more methods
which will briefly be discussed in later lectures,
by gravitational lensing
by pulsar timing

Problems

Problem 1 (10-20 min.)

1. The current precision in measurements of radial velocities by the Doppler
effect is 1m/s. Can a Jupiter like planet orbiting a star similar to the
Sun at a distance equal to the Sun-Jupiter distance in our solar system
be detected? (use www or other sources to find the necessary data)
2. What about an Earth like planet in orbit at a distance 1AU from the
same star?
3. Using the radial velocity method, is it easier to detect planets orbiting
closer or further away from the star?
4. In what distance range (from the mother star) does an Earth like planet
need to be in order to be detected with the radial velocity method?
(again use a star similar to the Sun). Compare with the distance SunMercury, the planet in our solar system which is closest to the Sun.
Problem 2 (20-30 min.) For stars which are sufficiently close to us,
their motion in the orbit about a common center of mass with a planet may
be detected by observing the motion of the star directly on the sky. A star
will typically move with a constant velocity in some given direction with
respect to the Sun. If the star has a planet it will also be wobbling up and
down (see figure 5). We will now study the necessary conditions which might
enable the observation of this effect.
11

Figure 5: The transversal wobbling of a nearby star due to its orbital motion
about the common center of mass with a planet. The angular extension of
the orbit is indicated by two small arrows.

1. The Hubble Space Telescope (HST) has a resolution of about 0.1 How
close to us does a star similar to the Sun with a Jupiter like planet (at
the distance of Jupiter) need to be in order for the HST to observe the
tangential wobbling of the star.
2. What about an Earth like planet at the distance of one AU from the
same star?
3. The closest star is Proxima Centauri at a distance of 4.22 l.y.. How
massive does a planet orbiting at the distance of 1 AU need to be in
order for the tangential wobbling of the star to be observed?
4. What about a planet at the distance of Jupiter?
5. If we can measure the tangential velocity component of a star, we can
get an estimate of the mass of the planet not only a lower limit. Show
that the mass of the planet can be expressed as
mp =

m2 P
2G

12

!1/3

vt

Figure 6: Velocity measurements of a star

(tangential velocity vt here is measured when the radial velocity is

zero)
Problem 3 (45 min. - 1 hour) In figure 6 we show observations of
the radial velocity of a star over a large period of time. We assume that
these data is a collection of data from several telescopes around the world.
Real data contain several additional complicated systematic effects which
are not included in this figure. For instance, changes in the velocity of the
Earth need to be corrected for in velocity measurements. Here we assume
that these corrections have already been made. Even if this plot does not
show you all the complications of real life, it does give an impression of how
data from observations may look like and how to use them to say something
about extrasolar planets. You see that this is not a smooth curve, several
systematic effects as for instance atmospheric instabilities give rise to what
we call noise.
1. This plot shows a curve with a wave like shape, can you explain the
shape of the curve?
2. Use this plot to give an estimate for the the perculiar velocity of the
star. Peculiar velocity is a term used to describe the average motion
of the star with respect to us, not taking into account oscillations from
planets.
13

Figure 7: The light curve of a star

3. Use the curve to find the maximum radial velocity vr of the star (with
respect to the center of mass) and the orbital period of the planet.
4. Spectroscopic measurements have shown the mass of the star to be 1.3
solar masses. Give an estimate of the lower bound for the mass of the
planet. The result should be given in Jupiter masses.
5. In figure 7 we show the light curve of the star for the same period of
time. Explain how this curve helps you to obtain the real mass of the
planet, not only the lower bound, and give an estimate of this mass.
6. In figure 8 we have zoomed in on a part of the light curve. Use the
figure to give a rough estimate of the density of the planet.
7. Is this a gas planet or a terrestrial planet?
Problem 4 (4-5 hours) At the following link you will find some files
containing simulated velocity and light curves of 10 stars:

http://folk.uio.no/frodekh/AST1100/lecture3/

14

Figure 8: The light curve of a star

Real data contain several additional complicated systematic effects which

are not included in these files. For instance, changes in the velocity of the
Earth need to be corrected for in velocity measurements. Here we assume
that these corrections have already been made. Even if these data do not
show you all the complications of real life, they will still give an impression
of how data from observations may look like and how to use them to say
something about extrasolar planets. Each file contains three rows, the first
row is the time of observation, counted in seconds from the first observation
which we define to be t = 0. We assume that these data is a collection of data
from several telescopes around the world, studying these stars intensively for
a given period of time (the length of this observing period is different for
each star). The second row gives the observed wavelength of a spectral
line (The H line) at 0 = 656.3nm in nm = 109 m. You need to use the
Doppler formula to obtain radial velocities yourself. You will see that this is
not a smooth curve, several systematic effects, i.e. atmospheric instabilities
give rise to what we call noise. As you will see, this noise makes exact
observations difficult. The third row shows the measured flux relative to the
maximum flux for the given star. Again, also these data contain noise.
Use Python, Matlab or other software/programming languages to solve
the following problems:
1. Estimate the peculiar velocity (the mean velocity of the star with re15

spect to Earth) for each of the 10 stars, taking the mean of the velocity
over all observations. Plot the velocity curves (subtract the mean velocity from the velocity for each observation) and light curves for the
ten stars. Which of the stars appear to have a planet orbiting? Which
of these planets are eclipsing their mother star?
2. The mass of the stars have been measured by other means, these are
0.8, 2.8, 0.5, 0.5, 1.8, 0.7, 1.6, 2.1, 7 and 8 solar masses for star 0-9
respectively. Can you, by looking at the velocity curves (velocity as a
function of time), find the lower limit for the mass of the planet for the
stars where you detected a planet. Find the numbers for the periods
and max radial velocities by eye.
3. If you, by looking at the light curve, discovered that some of the planets
are actually eclipsing the star, can you also estimate the radius and
density of these planets. Again, you will need to estimate the time of
eclipse by eye. Does any of the planets you have detected have the
possibilities for life (at least in the form that we know life) ?
4. You have made estimates of mass and radius using by-eye measurements. This is not the way that astrophysicists are working. Often,
advanced signal processing methods are employed in order to get the
best possible estimates. Also, scientific measurements always have uncertainties. The detailed methods for analyzing these data are outside
the scope of this course, but you will encounter this in more advanced
courses in astrophysics. Here we will show you a simple way to obtain
estimates which are more exact than the by-eye observations above.
A similar method will be used in other problems in this course. The
key to this method is the method of least squares. We will use this to
obtain more accurate periods and max radial velocities from the velocity measurements. We will model the velocity curves as cosine curves
in the following way,
vrmodel (t) = vr cos(2/P (t t0 )),

(4)

where vrmodel (t) is the theoretical model of the radial velocity as a function of time, vr is the maximal radial velocity, P is the period of revolution and t0 is some point for which the radial velocity is maximal (you
see that if t = t0 then the cosine term equals one). The unknown parameters in this model are vr , P and t0 . Only the two first parameters,
16

vr and P , are necessary in order to estimate the mass of the planet, but
we need to estimate all three in order to have consistent estimates of
the first two. We will now try to find a combination of these three parameters, such that equation (4) gives a good description of the data.
To do this, you need to write a computer code which calculates the
difference, or actually the square of the difference, between the data
and your model for a large number of values for the three parameters
t0 , P and vr . You need to define a function (an array in you computer)
(t0 , P, vr ) given as
(t0 , P, vr ) =

t=t
0 +P
X

(vrdata (t) vrmodel (t, t0 , P, vr ))2

t=t0

This function gives you the difference between the data and your model
for different values of t0 , P and vr . What you want to find is the function which best fits your data, that is, the model which gives the minimum difference between the data and your model. You simply want to
find for which parameters t0 , P and vr that the function (t0 , P, vr ) is
minimal. How do you find the parameters t0 , P and vr which minimizes
? In this case it is quite easy, try to follow these steps:
(a) Choose one of your stars which clearly has a planet orbiting.
(b) Look at your data: You know that for t = t0 , the velocity is
maximal. Look for the first peak in the curve and define a range
in time around this curve for which you think that the exact peak
must be. Define a minimum possible t0 and a maximum possible
t0 (being sure that exact peak is somewhere between these two
values). Then define a set of, say 20 (you choose what is more
convenient in each case) values of t0 which are equally spaced
between the minimum and maximum value.
(c) Do the same for vr , try to find a minimum and a maximum vr
which are such that you see by eye that the real exact vr is between
these two values. Then divide this range into about 20 equally
spaced values (maybe less depending on the case).
(d) Do the same thing for the period. Look at the time difference
between two peaks, and find a set of possible periods.
(e) Now, calculate the function for all these values of t0 , P and vr
which you have found to be possible values. Find which of these
17

about 203 combination of values which gave the smallest , thus

the smallest difference between data and model. These values are
now your best estimates of P and vr .
(f) Calculate the mass of the planet again with these values for P and
vr and compare with your previous by-eye estimates. How well
did you do in estimating by-eye ?
(g) Now repeat the procedure to estimate the exact mass for two other
stars with planets and compare again with your by-eye estimates.

18