ExtendedEssayFinalVersion 1
ExtendedEssayFinalVersion 1
Investigating the 21 cm
hydrogen line
The research question examined in this essay is “To what extent can the 21 cm
hydrogen line be used to reliably estimate the distance of its source from Earth?” The essay
examines and outlines the importance of one of the most important spectral lines in radio
astronomy known, the 21 cm hydrogen line. The aim of the essay is to explore the possibility
of mapping the source of this signal and to investigate the reliability of this method.
Data was collected by using a self-made radio telescope and measuring the signal in
several different regions of the sky. The data was then processed and compared to other data
gathered using an online radio astronomy database whose data is maintained using several
observing stations.
Several experimental imperfections were faced due to the basic equipment and
interference. Furthermore, the approach was also limited to bright sources that were relatively
near Earth and failed to explore sources farther away or less active ones. However, it was
concluded that the data gathered matched the data pertinent to the database to a large extent
(0.9981 correlation). Thus, it was concluded that the hydrogen line approach is indeed
1|Page
Acknowledgements
I would like to thank my English teachers Ivica Cikac and Irena Kocijan-Pevec for their
useful writing advices. I am hugely indebted to my father, Dusan Novak, for his help with the
equipment and assistance during the conduction of the measurements. Lastly, I am also
immensely grateful to my supervisor, Dinko Mestrovic, for his help, support and generous
2|Page
Table of contents
Introduction ........................................................................................................... 4
Research question............................................................................................... 5
Theory.................................................................................................................... 5
Observations ........................................................................................................ 12
Conclusion ........................................................................................................... 22
Bibliography ........................................................................................................ 25
Appendix ............................................................................................................. 26
3|Page
INTRODUCTION
The ancient Greek philosopher Plato once said: “Astronomy compels the soul to look
I have always been fascinated by astrophysics, the fusion of physics and astronomy.
However, astronomy is not just limited to the sense of sight, there is a whole field of
astronomy concerned with radio frequencies - radio astronomy. There are a vast number of
radio sources, such as stars and galaxies, in our universe which emit waves that we can detect
on Earth with radio antennas and telescopes. This fact opens up new possibilities of exploring
advancements in technology,
The subject matter of this paper is the 21 cm hydrogen line, one of the most important of
those radio signals coming from space. It is estimated that hydrogen makes up approximately
75% of our visible universe’s mass and monoatomic neutral hydrogen is the most common
form of it. Moreover, neutral hydrogen in the plane of the galaxy, in the form of giant
hydrogen clouds, emits a radio wave [4, 5]. Further, a small, basic radio telescope can be
4|Page
RESEARCH QUESTION
To what extent can the 21 cm hydrogen line be used to reliably estimate the distance of
Large hydrogen gas clouds in the plane of our galaxy are well known sources of the
hydrogen emission. The radio wave, with 21 cm wavelength and 1420.4058 MHz frequency,
can thus be received here on Earth [6]. Since there is relative motion present between the
source, cloud, and the Earth (Solar System), the frequency received was not the same as the
actual frequency of the cloud due to the Doppler effect. Based on that feature one can estimate
the velocity of the cloud along the line of sight, the so called radial velocity, and thus
calculate the distance of the cloud. This is very significant in astrophysics, because of the dust
and gas present in the galactic plane it is impossible to accurately estimate the distance of the
The aim of this research was to observe the hydrogen line in various regions of the sky
and based on the observations calculate the velocities and distances of the objects observed.
Furthermore, to question the accuracy of these numbers, the distances calculated using the
measurements have been compared to distances calculated using an online database with
THEORY
around it. Now, because of the very low temperatures in outer space the neutral hydrogen
atoms of the gas are in their lowest energy state, which means that the electron of an atom is
as close as it gets to the proton, nucleus. That said, both the electron and the proton spin in an
5|Page
individual axis, when both spin in the same direction that is called the parallel spin and
What actually causes the spectral line is the change from parallel to antiparallel spin.
Since the energy in an atom with parallel spin is greater than the energy in an atom with
antiparallel spin, when the spin of an atom changes, energy is emitted in the form of a low
energy photon causing the electromagnetic radiation spectral line. The wave emitted in this
case is a radio wave, i.e. low frequency (1420.4 MHz) and high wavelength (21 cm) [4, 6, 7].
PREPARATORY CALCULATIONS
Since the Earth (the Solar System) is not at rest, it is orbiting the Galactic Centre, the
signal received by the telescope will be Doppler shifted. Meaning it will contain the
movement of both the cloud and our Solar System around the Galactic Centre. Therefore one
can based upon this estimate whether the object observed is moving away from us or moving
towards us.
6|Page
R – the distance of the observed
object, hydrogen cloud, from the
Galactic Centre (v – its velocity)
l – galactic longitude
C – Galactic Centre
M – source (cloud)
T – tangential point
Figure 3: [7] Shows the visualized source and observer (Earth, Solar system)
The velocity observed by the radio telescope is the velocity in the line of sight,
illustrated below, the so called radial velocity. It can be expressed with the equation
𝑣𝑟 = 𝑣 cos 𝛼 − 𝑣0 sin 𝑐
This is the radial velocity and is expressed as the gas cloud’s velocity along the line of sight
The figure also indicates that 𝑐 = 𝑙 and that 𝛼 = 𝑎 . In addition, when looking at the
𝐶𝑇 = 𝑅 cos 𝛼 = 𝑅0 sin 𝑙
𝑅0 𝑅0
𝑣𝑟 = 𝑣 sin 𝑙 − 𝑣0 sin 𝑙 = (𝑣 − 𝑣0 ) sin 𝑙
𝑅 𝑅
7|Page
Assuming that the gas clouds in the Milky Way obey differential rotation, meaning that
circular velocity is constant with radius, since it is known that the velocity does not depend on
the radius (distance from the Centre) beyond a certain distance, one can write 𝑣 = 𝑣0 =
𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡. Thus expressing the distance of the gas from the Galactic Centre as
𝑅0 𝑣0 sin 𝑙
𝑅= (1)
𝑣0 sin 𝑙 + 𝑣𝑟
Figure 4: [12] The rotation curve of the Milky Way (observed and the
one that would be expected by applying Kepler’s laws)
The curve is horizontal beyond a certain distance, which means that the velocity does not
depend upon the radius after this distance any more. It can also be seen that the velocity is
significantly higher than Keplerian motion would suggest, which is attributed to dark matter,
Now, applying the cosine rule to the triangle CSM one can state that
8|Page
After applying the quadratic equation the distance of the cloud from the Sun, 𝑟 can be
expressed as [4, 5, 7]
So, it is in fact possible to come to two distinct solutions – one can obtain a negative
number and neglect it since it physically does not make sense, one can also obtain two
positive solutions. If 90° < 𝑙 < 270°, than cos 𝑙 < 0, so one obtains just one positive number.
In other cases where one gets two positive solutions it is advised to observe at the same
longitude, but different latitudes in order to determine the correct solution [7].
1) Antenna – this is where the signal and the telescope meet. The electromagnetic waves,
meaning the signal, are turned into electric currents, an oscillating electronic signal.
2) Receiver – the electronic signal then, through coaxial cabling, comes to this part.
Based on the voltage, the receiver is then able to compute the signal strength
It is also important to mention that most telescopes also have amplifiers in between the two
parts, because the signals observed is typically very weak. So, the electric currents are being
amplified.
For this investigation specifically, a basic, self-made radio telescope was constructed.
The instructions for this project were mainly from an article by M. Leech of Science Radio
Laboratories [3] in combination with other articles/instructions from the same author with
9|Page
Figure 5: [3] A basic sketch of how the equipment is connected into a telescope
1) Offset dish
A normal offset fed dish was used, because of financial reasons mostly. The dish is 95
cm tall and 85 cm wide, which is a bit small considering hydrogen line observations,
dimensions have been used in their case. The hardware of the dish also included a Ku-
Band LNB, which was helpful when placing the feed assembly.
A small feed assembly was constructed as an addition to the feed arm hardware, its
purpose is to maximize the gain of the feed. The feed was constructed by combining a
tin-plated coffee can with an aluminium adapter, to better illuminate the dish. The
LNA was soldered to the can. The assembly also functions like a filter, in that it
reduces the interference with signal whose frequency is less than 1100 MHz by the so
3) Amplifiers
There were three stages of amplification – firstly, a low noise amplifier (LNA) with
0.4 dB noise figure and 20 dB gain was used. Such a low noise figure is important for
10 | P a g e
this first stage, since the noise of the signal is extremely weak. Therefore a lower noise
figure increases the detection chances. The next two stages, to amplify the current
As for the connection between the motioned hardware, standard 75 ohm RG-6 coaxial
5) Receiver
A software defined radio (SDR) receiver was used. It was, on the one end, connected
to the computer and on the other, through cabling, to the power supply.
11 | P a g e
Figure 7: Shows the SDR used. This acted as the receiver and was
connected to the computer
OBSERVATIONS
Some well known radio sources that were observed such as the Cassiopeia constellation,
Sagittarius A and Taurus [8, 9]. Firstly, using the Stellarium application, the position and
galactic longitude of these objects were determined. The telescope beam was then pointed at
them when the signal was successfully targeted, it was recorded for approximately 15 to 20
GnuRadio SDR framework for Linux was used for this investigation – two components
frequency graph and simple_ra (simple radio astronomy) which is used to not only record the
data, but also it can be used to create power vs velocity graphs which were useful later for the
analysis.
When searching for a signal, osmocom_fft component was used. When opening the simple_ra
to record, however, some parameters must have been entered: the central frequency
(1420.4058 MHz), the gain (dB), the sample rate (1 million Hz), the logging-rate (5 s –
12 | P a g e
suggested value, sets the logging interval for logging data), the detector bandwidth (around
Central axis
The figure above shows an example of a hydrogen observation. It shows the Spectral
window, one out of four windows of the simple_ra application. The example is from
observing the Cassiopeia constellation. It can be seen that the peak is right to the central
frequency, meaning that the apparent frequency, the one we observe, is larger than the known
accepted value of the hydrogen emission frequency. So, considering the Doppler effect, this
means that the object, cloud, observed is moving towards the Earth in the example. It can also
be seen that there is a lot of interference with the very weak hydrogen signal.
In order to get the velocities of the objects observed, the simple_ra data was processed using
the spectral data utility. What this utility plots are graphs with regards to velocity, not MHz,
which is how professionals observe the galactic plane. The utility also takes the Doppler shift
13 | P a g e
into account when constructing these plots, using the apparent and central value of the
∆𝑓 𝑣
= −
𝑓0 𝑐
Where the frequency shift, ∆𝑓 = 𝑓 − 𝑓0 , where 𝑓 is the observed frequency and 𝑓0 the rest
Figure 9: A processed velocity graph from which the velocity was read
The figure above shows one of these graphs. This particular example shows the graph
of the Taurus region. Notice that there are multiple peaks shown. Each of these peaks is a
cloud, meaning that the in the example there were 2 separate clouds present in the field
observed. Note that other, smaller peaks were ruled out as disturbances, interference.
Consequently, the velocity of each cloud was noted. Also note that there are also negative
14 | P a g e
velocities present, this means that the object observed in approaching (Doppler effect).
telescope using the Stellarium computer application, the uncertainty for the calculations is
estimated to be 1˚. In the figure below, from which one can easily identify region of 10˚ (one
square), I estimate that my precision at pointing towards the certain source was about 1˚,
Figure 10: An example of a target region of the Milky Way, Cassiopeia in the Stellarium application, the
galactic grid can also be seen. Note that the approximate region spans over 2˚ longitude (do not take into
account the latitude)
As for the velocity, the simple_ra program creates picture (png) plots and the
uncertainty of the velocities was given in terms of picture resolution, 0.103 km/s (Fig. 9, same
for every picture). However, the pictures had to be analysed using a different program in order
15 | P a g e
to maintain the raw data. I analysed them using the g3data program. When a picture graph is
to be analysed in this program, one has to select the 𝑦 and 𝑥 boundaries oneself with a mouse
click and then, when looking for the velocity value of the peak, I had to move the mouse to
this point.
Given that further uncertainty arises from my precision to point the mentioned things
correctly, I estimate the uncertainty of the measurements has to therefore be increased by 0.01
km/s, so 0.113 km/s. Note that this figure was used in the calculations, the value below is
RAW DATA1
As described above, a longitude represents a certain region of the sky targeted and there
can be multiple peaks in one region (Fig. 9) representing different clouds with different
1
See Appendix for unrounded values. Same for Table 2 and 3
16 | P a g e
Longitude, l (˚) ± 1˚ Radial velocity (km/s) ± 0.1 km/s
105 -4.3
-6.1
-14.0
126 -6.5
-4.5
-2.2
170 -6.3
-4.5
12 2.6
110 -2.4
-38.6
Table 1: The longitudes of the observations done and the peaks, meaning radial
velocities of separate clouds measured for each longitude (region)
17 | P a g e
Table 3
Tables 2 and 3: The distances R and r calculated for each velocity (cloud)
Both the table are based upon Eq. (1) and Eq. (2) from the theory chapter.
𝑚𝑎𝑥−𝑚𝑖𝑛
The uncertainty for each was calculated using the same principle; . Therefore,
2
Regardless of the constants, when looking at the numerator of the maximum value, one can
see that the longitude in the sines is minimal, and maximal for the numerator of the minimal
value. This is because nearly all the observed values are between 90 and 180 degrees, so the
lower the longitude, the greater the sine and vice versa. Now, the denominator has to be
minimal for the maximum value and vice versa, so the minimal or maximal radial velocity
values were used based on the uncertainty discussed above. The exact opposite was done
The uncertainty for 𝑟, on the other hand, using Eq. (2) was calculated by
18 | P a g e
1
∆𝑟 = (√𝑅𝑚𝑎𝑥 2 − 𝑅0 2 sin2 𝑙𝑚𝑖𝑛 − √𝑅𝑚𝑖𝑛 2 − 𝑅0 2 sin2 𝑙𝑚𝑎𝑥 + 𝑅0 (cos 𝑙𝑚𝑎𝑥 −
2
(4)
cos 𝑙𝑚𝑖𝑛 ))
The same as above applies for the min and max sine values, however, the cosine values
increase as longitude increases for the given values. So, greater longitudes for maximum and
Again, the opposite was done for the data at 12˚. It was also estimated that the observation
done at approximately 12˚ was the Sagittarius A star cloud, a large well known radio source
right at the centre of the Milky Way [8, 13], so the longitude for further calculations was
estimated to be 12˚ 32’ 59.8’’ (the figure was searched for the mentioned cloud specifically
using Stellarium), instead of 12˚, but with the uncertainty being the same, 1˚.
For instance taking the value at 105˚, the cloud with radial velocity -4.3 km/s, the
Now, the distance 𝑟 is calculated as: ±√(2.57 × 1017 )2 − (2.62 × 1017 )2 sin2 105 ˚ +
2.62 × 1017 cos 105˚ = 2.67 𝑜𝑟 9.96 × 1016 , now eliminating the latter solution, because of
1
± 2 (√(2.61 × 1017 )2 − (2.62 × 1017 )2 sin2 104˚ −
√(2.53 × 1017 )2 − (2.62 × 1017 )2 sin2 106˚ + (2.62 × 1017 )(cos 106˚ − cos 104˚)) =
±1.94 × 1017 𝑘𝑚
19 | P a g e
Also note that the uncertainty values in the tables above have been rounded up to one
COMPARISON ANALYSIS
In order to evaluate the accuracy of the data gathered by the measurements, I used
enables the user to simulate radio observations or to use their archives as a resource.
Galactic coordinates
Peak (cloud)
Even though the galactic latitude did not matter for my calculations as stated above, it was
necessary here to search the database. Therefore, it was, like the longitude, estimated using
the Stellarium application. Note that I also estimated that the uncertainty for the latitude is 1˚
and thus when searching the database this was taken into account. So, the values that were
20 | P a g e
most similar to the values based on the measurements were taken for further analysis. Also
note that there is no uncertainty given for the values obtained from the database. However, it
1E+18
Distance of each cloud (database and measured)
8E+17
Distances from Earth, r (km)
6E+17
4E+17
2E+17
0
105˚ 126˚ 170˚ 18˚
12˚ 110˚
-2E+17
-4E+17
Longitudes of the observed sources (˚)
Figure 13: The distances of the clouds calculated using my measurements (blue) plotted with the
distances calculated using the database values (red)
Almost all the values obtained from the database can be assigned to one or even more of the
values I obtained, with respect to the error bars. When looking at the clouds observed at 126˚,
21 | P a g e
it can be seen that the value I wasn’t able to detect was way further from Earth than the other
sources.
Figure 14: The correlation of the database values and the measured ones.
Created using LoggerPro3.12
Taking the values that were calculated using the database and plotting them against the
values calculated using the measurements that were most likely coming from the same source
(minimal difference between the two values in the case with the observation at 126˚, since
more values fit the database number), one can see a very strong positive correlation, 0.9981.
CONCLUSION
Overall, the project was successful in that it showed that radio waves, the hydrogen
emission in particular, can be used to estimate the velocities and furthermore the distance of
celestial objects reliably. The distances calculated using the measurements were in line with
the distances calculated using the database values, within reasonable error. Furthermore,
22 | P a g e
judging from the correlation in Fig. 14, one can say that these indeed originated from the
same source.
However, some flaws have to be pointed out as well, for instance the fact that some
values were not assigned to a database value. This can be linked to the equipment – when
looking at database values that turned out to come from a source very far away, one can say
that it was not possible to detect because of the basic equipment (Fig. 13 at 126˚). Due to this
fact, the data collection was limited to relatively near and brighter sources which reduced the
Similarly, it can be seen that values farther away seem follow the correlation pattern in
Fig. 14 a little less. This leads to the conclusion that the distances calculated for farther away
sources are less reliable, which can, again, be linked to the limitations associated with the
imperfections of the equipment. Another thing to notice is that when looking at Fig. 13, it
seems as though multiple of the measurements obtained could be linked to the same database
value, which leads me to the conclusion that perhaps they originated from the same source
(cloud) and due to some signal interference (discussed below) the software recognised them
instance, in order to get a sense of how weak of a signal the telescope can detect, some
regions that are known to be more “quiet”, but still having visible hydrogen sources like the
Acquarius cold spot [8] were tried to be targeted on several different occasions. However,
these attempts were without any success. So, the data collection was very limited to “louder”
sources.
23 | P a g e
One of the imperfections of the equipment was, for instance, the feed assembly. Even
though the can in the feed assembly was made of galvanized steel, an aluminium adapter was
used to illuminate the dish instead of a galvanized steel adapter [3]. This change can influence
the way the signal was illuminated and thus make the measurements less accurate or one
could argue that this influenced the fact that the telescope was not able to detect weaker
Furthermore, another big threat for the accuracy of the measurements was the
interference. Even though the measurements were, on purpose, not done in an urban area, they
away. This is most likely the biggest cause of the radio waves to interfere and thus make the
data less reliable, similar occurrences have been noted by other amateur astronomers [1]. I can
only estimate the operating frequency, which I would say is about 1100-1300 MHz. Now, this
signal can be heard sometimes at the place of the observations and more often than not it is
very weak. However, it can be said that it can cause a systematic error to a certain extent in
some measurements, since it operated at a frequency too big to be attenuated by the feed
assembly.
Another thing to consider is the structure of the dish. An offset-fed dish was used for
this investigation. However, using a prime focus dish would be a significant improvement for
further studies. Not only is a prime focus dish easier to point, but also its gain is proportional
to its diameter – it is ideal to use for a relatively short radio wavelength, such as the 21 cm
signal [2]. Furthermore, this is the kind of dish that professionals use for these very reasons.
24 | P a g e
BIBLIOGRAPHY
[1] M. Leech and D. Ocame, "A year of Gnu Radio and SDR astrononomy: experience,
[3] M. Leech, Science Radio Laboratories, Inc, "A 21cm Radio Telescope for the Cost-
Conscious" 2006.
2016].
[7] C. Horellou and D. Johansson, "Hands-On Radio Astronomy, Mapping the Milky
25 | P a g e
[10] “Parkes Observatory,” Parkes Observatory. [Online]. Available:
https://www.csiro.au/en/Research/Facilities/ATNF/Parkes-radio-telescope/Parkes-
2016].
[12] J. Schaff, “The Nature of Space and of Gravitation,” Journal of Modern Physics, 01-
[Accessed: 28-Aug-2016].
APPENDIX
The tables with the values used for calculations (all figures shown are to be multiplied by
1017):
Table 1
105 -4.261
-6.103
-13.984
126 -6.479
-4.532
-2.173
170 -6.341
-4.530
12 2.590
110 -2.420
-38.593
26 | P a g e
Table 2
Table 3
27 | P a g e