University of Ghana http://ugspace.ug.edu.gh
UNIVERSITY OF GHANA
COLLEGE OF BASIC AND APPLIED SCIENCES
MONITORING AND HIGH RESOLUTION STUDIES OF
SOME HYDROXYL AND METHANOL MASERS IN THE
HIGH MASS STAR FORMING REGION G9.62 + 0.20 E
BENEDICTA WOODE
DEPARTMENT OF PHYSICS
JULY 2020
 University of Ghana http://ugspace.ug.edu.gh
UNIVERSITY OF GHANA COLLEGE OF BASIC AND
APPLIED SCIENCES
MONITORING AND HIGH RESOLUTION STUDIES OF
SOME HYDROXYL AND METHANOL MASERS IN THE
HIGH MASS STAR FORMING REGION G9.62 + 0.20 E
BY
BENEDICTA WOODE
(10220100)
A THESIS SUBMITTED TO THE UNIVERSITY OF
GHANA, LEGON, IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE AWARD OF THE DEGREE
OF DOCTOR OF PHILOSOPHY IN PHYSICS
DEPARTMENT OF PHYSICS
JULY 2020
 University of Ghana http://ugspace.ug.edu.gh
ABSTRACT
This thesis presents the observational data reduction and results of an investigation into the
1.6 GHz OH and 6.7 GHz methanol masers in the G9.62+ 0.20E massive star forming re-
gion. Monitoring observations from the Hartebeesthoek Radio Astronomy Observatory (Har-
tRAO) radio telescope were employed to establish correlated variabilities between the 6.7 GHz
methanol masers and the left and right circular polarizations (LCP and RCP) of the mainline
1.6 GHz OH masers. This is the first reported incidence of periodic variations of the LCP and
RCP components of the 1.6 GHz OH mainline masers associated with the 6.7 GHz methanol
masers in this high mass star forming region. Correlated variability between the varying fea-
tures in the LCP components of both hydroxyl masers and methanol masers was observed but
the variability profiles were different for both species. Time lags were recorded between the
methanol and the OH maser variabilties. Within the OH masers the variability profiles were
preceded by a dip in flux density. The first high resolution observation of the L-band and C-
band continuum of G9.62+ 0.20E were conducted with the e-MERLIN interferometric array.
A flux density of 2.25±0.20 mJy and an angular size of 130 mas were measured for the C-band
continuum. At L-band, an upper limit of 162 µJy on the integrated flux density was calculated
at the position of the continuum. These results unlike previous observations reveal that the
spectrum turns over steeply at low frequencies which could be explained with a truncated in-
verse square law density distribution model. The peak brightness temperature indicates that
the continuum is optically thin at C-band and optically thick at L-band. In the context of the
colliding wind binary model for periodic maser sources it is the variation of the seed photons
from the background HII region that explains the variability. The continuum results in this
thesis place firm constraints on models of the HII region that can be used to test the predictions
of the colliding wind binary model in terms of the variability profiles of the OH and methanol
masers.
i
 University of Ghana http://ugspace.ug.edu.gh
DECLARATION
I, Benedicta Woode, hereby declare that except where specific reference is made to the work
of others, the contents of this dissertation are original and have not been submitted in whole or
in part for consideration for any other degree or qualification in this or any other university and
that this submission undertaken in the University of Ghana is my own work under the supervi-
sion of the undersigned.
.............................................
Candidate: Benedicta Woode
Supervisors:
............................................
Dr. Amos Kuditcher
...............................................
Prof. Melvin G. Hoare
..............................................
Dr. Sharmila Goedhart
ii
 University of Ghana http://ugspace.ug.edu.gh
DEDICATION
To the memory of my maternal grandmother, Madam Georgina Adjoa Besema Enchill, and
to astronomy in Ghana.
iii
 University of Ghana http://ugspace.ug.edu.gh
ACKNOWLEDGEMENTS
This work is the output of four years of research which several persons have through techni-
cal support, advice, financial support and in diverse ways helped to produce. I wish to express
my sincere gratitude to all of them.
I wish to thank Dr. Amos Kuditcher for guiding and nurturing me from my undergraduate
days till now, Prof. Melvin Hoare, the man who makes everything happen, and Dr. Sharmila
Goedhart for giving me the data on her baby, G9.62+0.20 E. I am also grateful to Dr. Gordon
Macleod for scheduling my HartRAO observations.
I wish to acknowledge support from Dr. Javier Molden for painstakingly taking me through
the rudiments of interferometric data reduction, Dr. James Chibueze, Dr. Jabulani Maswang-
egy, Mavis Seidu, Dr. Benito Marcote and Ziad Modak who were very instrumental in my
study of Python. I also thank Dr. Rosalind Skelton who introduced me to the rudiments of the
field and Dr. Robert Beswick who has been as he himself will put it ’lovingly excellent.’ I am
also grateful to Prof Alberto Sanna for providing the Q-band size of G9.62+0.20 E.
It has been a pleasure working with members of the Department of Physics in the University
of Ghana, School of Physics and Astronomy in the University of Leeds, Jodrell Bank Center
for Astronomy in the University of Manchester and the Ghana Space Science and Technology
Institute; I am grateful for their support.
Research reported in this thesis was supported by a Newton Fund project, DARA (Devel-
opment in Africa with Radio Astronomy), and awarded by the UK’s Science and Technology
Facilities Council (STFC) - grant reference ST/R001103/1. I am grateful to Prof. Yaa Nti-
amoah Baidoo and the Carnegie Corporation, New York, for funding this work under the Next
Generation of Academics in Africa grants. I am also indebted to the European Union Horizon
2020 Research and Innovation programme via Jumping Jive for funding my stay in Manchester.
I thank my family and friends for the love and encouragement throughout my long years of
study. And to my husband, Akwasi Boakye and my son Nana Kwabena Baffour Boakye, what
exactly could I have done without you?
iv
 University of Ghana http://ugspace.ug.edu.gh
TABLE OF CONTENTS
ABSTRACT i
DECLARATION ii
DEDICATION iii
ACKNOWLEDGEMENTS iv
LIST OF FIGURES ix
LIST OF TABLES xxii
LIST OF SYMBOLS xxv
LIST OF ABBREVIATIONS xxvi
1 INTRODUCTION 1
1.1 HII regions in massive star formation . . . . . . . . . . . . . . . . . . . . . . . 11
1.2 HII regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.1 Free-free emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.2 Classification of HII regions . . . . . . . . . . . . . . . . . . . . . . . 14
1.2.3 Molecule formation in the interstellar medium . . . . . . . . . . . . . . 14
1.3 Astrophysical masers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.3.1 Maser theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.4 Variability of masers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
1.5 G9.62+0.20E . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
1.6 Periodicity of the 6.7 GHz methanol masers in G9.62+0.20E . . . . . . . . . 31
v
 University of Ghana http://ugspace.ug.edu.gh
1.6.1 Pulsating star models . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.6.2 Accretion disk models . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.6.3 Colliding wind binary model . . . . . . . . . . . . . . . . . . . . . . . 33
1.7 Thesis goals and overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2 MONITORING THE POLARISATIONS OF THE HYDROXYL MASERS IN
THE PERIODIC METHANOL MASER SOURCE G9.62+0.20E 37
2.1 Observational techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.1.1 Frequency switching . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.1.2 Drift scans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.2 Data reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.2.1 Bandpass correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.2.2 Radio-frequency interference excision . . . . . . . . . . . . . . . . . . 46
2.2.3 Smoothing of spectrum . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.2.4 Baseline flattening . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.2.5 Pointing correction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.2.6 Flux density calibration . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.3.1 Averaged spectra for regular monitoring . . . . . . . . . . . . . . . . . 57
2.3.2 Averaged spectra for intensive monitoring . . . . . . . . . . . . . . . . 61
2.3.3 Spectra from Caswell et al. (2013) . . . . . . . . . . . . . . . . . . . . 64
2.3.4 Comparison of the averaged spectra of OH for regular and intensive
monitoring window with Caswell et al. (2013) . . . . . . . . . . . . . . 66
2.3.5 Comparison of the averaged spectra of methanol for the regular and
intensive monitoring window with Goedhart et al. (2004) . . . . . . . . 67
2.3.6 Range of variation for regular monitoring . . . . . . . . . . . . . . . . 68
2.3.7 Range of variation for intensive monitoring . . . . . . . . . . . . . . . 73
2.3.8 Comparison of the range of variation for regular and intensive monitor-
ing windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
2.3.9 Time series of some selected channels: regular monitoring . . . . . . . 79
vi
 University of Ghana http://ugspace.ug.edu.gh
2.3.10 Times series of some selected channels: intensive monitoring . . . . . . 84
2.3.11 Regular monitoring of some selected channels in all transitions . . . . . 89
2.3.12 Intensive monitoring of some selected channels in all transitions . . . . 92
2.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3 HIGH RESOLUTION STUDIES OF THE 1.6 GHz HYDROXYL MASER SPOTS
AND THE L-BAND RADIO CONTINUUM IN G9.62+0.20E 97
3.1 The e-MERLIN interferometric array . . . . . . . . . . . . . . . . . . . . . . . 98
3.2 Interferometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
3.2.1 Aperture synthesis and imaging . . . . . . . . . . . . . . . . . . . . . 101
3.2.2 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
3.3 Observation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
3.4 Data reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
3.4.1 Pre-processing data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
3.4.2 Continuum calibration . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.4.3 Self calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
3.4.4 Spectral line calibration . . . . . . . . . . . . . . . . . . . . . . . . . . 125
3.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
3.5.1 The L-band continuum . . . . . . . . . . . . . . . . . . . . . . . . . . 132
3.5.2 Maser spots in the LCP and RCP components at 1.665 GHz and 1.667 GHz137
3.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
4 HIGH RESOLUTION STUDIES OF THE 6.7 GHz METHANOL MASER SPOTS
AND THE C-BAND RADIO CONTINUUM IN G9.62+0.20E 151
4.1 Observation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
4.2 Data reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
4.2.1 Preprocessing data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
4.2.2 Continuum calibration . . . . . . . . . . . . . . . . . . . . . . . . . . 154
4.2.3 Spectral line calibration . . . . . . . . . . . . . . . . . . . . . . . . . . 158
4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
vii
 University of Ghana http://ugspace.ug.edu.gh
4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
5 CONCLUSION AND FUTURE RESEARCH 178
5.1 Improving observational constraints . . . . . . . . . . . . . . . . . . . . . . . 180
REFERENCES 182
viii
 University of Ghana http://ugspace.ug.edu.gh
LIST OF FIGURES
Figure1.1 A three-colour composite Spitzer image of the G9.62+ 0.20 complex
showing emissions at 8 µm in red, 4.5 µm in green, and 3.6 µm in blue. The
contours are constant intensity lines for which the white contours trace dust
emissions at 450 µm and the red contours trace photo-dissociation regions at
8 µm (Liu et al., 2017). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Figure1.2 The four stages of the low mass star formation process as illustrated by
Shu et al. (1987); van Dishoeck and Blake (1998); Greene (2001). Plate (a)
depicts the fragmentation of a molecular cloud into pre-stellar cores as the cloud
collapses under its gravity. In plate (b) the proto-star builds up mass from in-
falling gas as it continues to collapse. Plate (c) shows the formation of a disc in
the nebula as the proto-star conserves angular momentum and the breakout out
of bipolar outflows along the the axis of rotation of the T-Tauri star. Finally the
nebula is swept away in Plate (d) revealing the newly formed pre-main sequence
star and the remains of the accretion disk. . . . . . . . . . . . . . . . . . . . . 7
Figure1.3 The observed evolutionary stages of high mass stars after the fragmenta-
tion and collapse of molecular clouds (i). The prestellar cold collapsing cores
(ii) develop into hot molecular cores (iii), which are heated by the embedded
massive star. An ultracompact HII region (region of ionised gas) is formed
around the young star (iv) which evolves into a diffuse HII region in OB asso-
ciations (v) (Purcell et al., 2006; Breen et al., 2010). . . . . . . . . . . . . . . . 10
ix
 University of Ghana http://ugspace.ug.edu.gh
Figure1.4 The ideal radio spectrum of an HII region. The solid line shows the
free-free emission from the UCHII region and the dashed line depicts free-free
emission from the HCHII region. The vertical dotted line separates the optically
thick part from the optically thin part of the UCHII region (Kurtz, 2005). . . . . 13
Figure1.5 Model of the hydroxyl molecule. . . . . . . . . . . . . . . . . . . . . . 15
Figure1.6 Model of the methanol molecule. . . . . . . . . . . . . . . . . . . . . . 16
Figure1.7 An excited atom de-exciting from a high energy level to a low energy
level via the spontaneous emission of a photon. . . . . . . . . . . . . . . . . . 19
Figure1.8 An incoming photon of a specific frequency interacts with an excited
atom, causing it to de-excite to a lower energy level. The emitted photon is
identical to the incident photon. . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Figure1.9 The different processes that populate or depopulate the levels in a col-
lection of two level systems in thermodynamic equilibrium. The radiative pro-
cesses are absorption characterized by coefficient B12, spontaneous emission
characterized by coefficient A21 and stimulated emission characterized by co-
efficient B21. Collisional processes are characterized by coefficients C12 for
excitation and C21 for de-excitation. . . . . . . . . . . . . . . . . . . . . . . . 20
Figure1.10 Energy levels of the two ladders of the rotational states of OH. The
splitting of the ground state of OH occurs via lambda doubling and hyperfine
interactions. The broken lines show the pumping scheme for the 1612 MHz
maser transition in OH (Elitzur, 1976). . . . . . . . . . . . . . . . . . . . . . . 23
Figure1.11 Part of the rotational energy level diagram of the hydroxyl ground state
showing the hyperfine splitting of the rotational states that give rise to the 18 cm
hydroxyl emission. Shown are the four hyperfine masing transitions of the OH
ground state. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
x
 University of Ghana http://ugspace.ug.edu.gh
Figure1.12 Energy level diagram for E-type (left) and A-type (right) symmetry states
of the ground state methanol transitions. The E-type symmetry for which k < 0
indicates transitions for CH3OH E− levels whereas the E-type symmetry for
which k > 0 indicates transitions for CH OH E+3 levels. The A-type symme-
try for which k > 0 indicate both CH −3OH A and CH OH A+3 levels. The A-
type levels are further split into doublets indicated by + and symbols. The ar-
rows in both A-type and E-type diagrams show observed methanol maser tran-
sitions. The arrow in the A-type diagram shows the 6.7 GHz maser transition
(De Buizer, 2000). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Figure1.13 Maser saturation plot adopted from Cohen (1989) in which for an unsat-
urated system the maser growth is exponential until such a distance demarcated
with the broken line when it becomes saturated and the growth becomes linear. 28
Figure1.14 The Zeeman splitting of the hydroxyl 2π3/2 hyperfine levels that gives
rise to the 1665 MHz transition polarizations. Adapted from Garcia-Barreto
et al. (1988). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Figure1.15 Time series of the peak velocity channel of the 6.7 GHz masers in G9.62+
0.20E (Goedhart et al., 2004). . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Figure1.16 A toy model of the colliding wind binary model (van der Walt, 2011a). . 33
Figure2.1 The raw total power spectrum from the 1024 channel spectrometer show-
ing the bandpass and the maser signal in the left circular polarisation (LCP). . . 42
Figure2.2 The second spectrum of the frequency switched pair in LCP showing
the bandpass with the spectrometer artefact at about −7 km/s and maser line
emission offset by half the bandwidth to the right of the spectrum. . . . . . . . 43
Figure2.3 The first observation of the frequency switched pair and the second ob-
servation of the frequency switched pair both showing positive and negative
velocity maser features offset by 22.5 km/s. . . . . . . . . . . . . . . . . . . . 44
Figure2.4 Combined observation of the frequency switched pair. . . . . . . . . . . 45
Figure2.5 Averaged frequency switched spectra showing the baseline level removed. 46
xi
 University of Ghana http://ugspace.ug.edu.gh
Figure2.6 Spectrum showing persistent radio-frequency interference manifesting
as spikes and dips. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Figure2.7 The spectrum in Figure 2.6 with RFI spikes and dips removed. . . . . . . 48
Figure2.8 Ringing in the baseline smoothed out with a Hanning window algorithm. 49
Figure2.9 A fourth-order polynomial fit of a green curve to the spectral baseline
showing an uneven fit to the bandpass. The blue coloured part of the spectrum
shows the masked maser line and the orange part of the spectrum shows the
baseline. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Figure2.10 The top plot illustrates a third-order polynomial fit of a green curve to the
spectral baseline with the masked range being the start and end velocity of line
emissions shown in blue.The bottom plot shows the results of the subtraction
of the baseline fit from which flattens the bandpass. . . . . . . . . . . . . . . . 50
Figure2.11 The spectrum showing the flattened baseline of both LCP and RCP with
the mean signal at 0 K antenna temperature. . . . . . . . . . . . . . . . . . . . 51
Figure2.12 Peak power of the spectra taken at the half power point of cardinal points
(north, south, east and west) and on the source center. . . . . . . . . . . . . . . 52
Figure2.13 The left plot shows the north-south pointing offset with an offset of
0.015deg from the center and the right plot displays the east-west pointing
offset of 0.017deg from the center. . . . . . . . . . . . . . . . . . . . . . . . . 53
Figure2.14 Fitting of a Gaussian beam plus linear baseline to the drift scan on Virgo
A to calculate the amplitude of the peak. . . . . . . . . . . . . . . . . . . . . . 55
Figure2.15 Calibrated spectrum after correction factor PSS has been applied to the
on-source amplitude of the antenna temperature of the hydroxyl spectrum to
covert it to flux density. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Figure2.16 Averaged spectrum for the 1665 MHz maser line in LCP. . . . . . . . . . 58
Figure2.17 Averaged spectrum for the 1665 MHz maser line in RCP. . . . . . . . . . 58
Figure2.18 Averaged spectrum for the 1667 MHz maser line in LCP. . . . . . . . . . 59
Figure2.19 Averaged spectrum for the 1667 MHz maser line in RCP. . . . . . . . . . 60
Figure2.20 Averaged total intensity spectrum for the 6668 MHz maser line. . . . . . 60
xii
 University of Ghana http://ugspace.ug.edu.gh
Figure2.21 Averaged spectrum for the 1665 MHz maser line in LCP. . . . . . . . . . 62
Figure2.22 Averaged spectrum for the 1665 MHz maser line in RCP. . . . . . . . . . 62
Figure2.23 Averaged spectrum for the 1667 MHz maser line in LCP. . . . . . . . . . 63
Figure2.24 Averaged spectrum for the 1667 MHz maser line in RCP. . . . . . . . . . 63
Figure2.25 Averaged total intensity spectrum for the 6668 MHz maser line. . . . . . 64
Figure2.26 The 1665 MHz spectrum of the three regions G9.619+0.193, G9.620+0.194
and G9.621+0.196 in G9.62+0.20 from Caswell et al. (2013) showing the left
circular polarisation, right circular polarisation and linear polarisation features. 65
Figure2.27 The 1667 MHz spectrum of the three regions G9.619+0.193, G9.620+0.194
and G9.621+0.196 in G9.62+0.20 from Caswell et al. (2013) showing the left
circular polarisation, right circular polarisation and linear polarisation features. 65
Figure2.28 Range of variation of flux density in the spectrum of 1665 MHz LCP.
The blue line represents the upper envelope, the green line shows the averaged
spectrum, and the orange line shows the lower envelope. . . . . . . . . . . . . 69
Figure2.29 Range of variation of flux density in the spectrum of 1665 MHz RCP.
The blue line represents the upper envelope, the green line shows the averaged
spectrum, and the orange line shows the lower envelope. . . . . . . . . . . . . 70
Figure2.30 Range of variation of flux density in the spectrum of 1667 MHz LCP.
The blue line represents the upper envelope, the green line shows the averaged
spectrum, and the orange line depicts the lower envelope. . . . . . . . . . . . . 71
Figure2.31 Range of variation of flux density in the spectrum of 1667 MHz RCP.
The blue line represents the upper envelope, the green line shows the averaged
spectrum, and the orange line depicts the lower envelope. . . . . . . . . . . . . 72
Figure2.32 Range of variation of flux density in the spectrum of 6668 MHz. The blue
line represents the upper envelope, the green line shows the averaged spectrum,
and the orange line depicts the lower envelope. . . . . . . . . . . . . . . . . . . 73
Figure2.33 Range of variation of flux density in the spectrum of 1665 MHz LCP.
The blue line represents the upper envelope, the green line shows the averaged
spectrum, and the orange line depicts the lower envelope. . . . . . . . . . . . . 74
xiii
 University of Ghana http://ugspace.ug.edu.gh
Figure2.34 Range of variation of flux density in the spectrum of 1665 MHz RCP.
The blue line represents the upper envelope, the green line shows the averaged
spectrum, and the orange line depicts the lower envelope. . . . . . . . . . . . . 75
Figure2.35 Range of variation of flux density in the spectrum of 1667 MHz LCP.
The blue line represents the upper envelope, the green line shows the averaged
spectrum, and the orange line depicts the lower envelope. . . . . . . . . . . . . 76
Figure2.36 Range of variation of flux density in the spectrum of 1667 MHz RCP.
The blue line represents the upper envelope, the green line shows the averaged
spectrum, and the orange line depicts the lower envelope. . . . . . . . . . . . . 77
Figure2.37 Range of variation of flux density in the spectrum of 6668 MHz. The blue
line represents the upper envelope, the green line shows the averaged spectrum,
and the orange line depicts the lower envelope. . . . . . . . . . . . . . . . . . . 78
Figure2.38 Time series for selected velocity channels of the 1665 MHz LCP com-
ponent in the regular monitoring window. The features in the velocity channels
1.36 km/s, 1.45 km/s and 1.53 km/s follow a monotonic increase in flux den-
sity from 6 Jy to 18 Jy whilst the features in the velocity channels 9.62 km/s
and 9.84 km/s show no variation. The black + signs plotted on the 1.36 km/s
velocity channel show the times the flares are expected. . . . . . . . . . . . . . 80
Figure2.39 Time series for selected velocity channels of the 1665 MHz RCP com-
ponent in the regular monitoring window. The features in the velocity channels
1.66 km/s, 1.71 km/s, 1.75 km/s, 2.72 km/s and 2.94 km/s only show spo-
radic variation in flux density from 4 Jy to 7 Jy. The black + signs plotted on
the 1.66 km/s velocity channel show the times the flares are expected. . . . . . 81
Figure2.40 Time series for selected velocity channels of the 1667 MHz LCP com-
ponent in the regular monitoring window. Features in the velocity channels
1.58 km/s, 1.62 km/s and 1.67 km/s vary in flux density from 6 Jy to 24 Jy
with two significant flares at 56900 MJD and 57100 MJD marked with + signs
on channel 1.58 km/s. The features in the velocity channels 2.15 km/s and
3.11 km/s show no change in flux density except a minor flare at 56900 MJD. . 82
xiv
 University of Ghana http://ugspace.ug.edu.gh
Figure2.41 Time series for selected velocity channels of the 1667 MHz RCP com-
ponent in the regular monitoring window. The features in the velocity channels
1.58 km/s, 1.62 km/s, 1.67 km/s, −0.92 km/s and −0.42 km/s show no vari-
ation. The two black + signs plotted on the 1.58 km/s velocity channel show
the times the flares are expected. . . . . . . . . . . . . . . . . . . . . . . . . . 83
Figure2.42 Time series for selected velocity channels of the 6668 MHz methanol
maser transition in the regular monitoring window. The features at 1.18 km/s,
1.23 km/s and 1.27 km/s show 15 periodic flares. Part of the data from
54900 MJD to 55600 MJD has been flagged because of bad data. However
the features in channel −0.97 km/s and channel −0.92 km/s show a mono-
tonic increase in flux from 50 Jy to 150 Jy. . . . . . . . . . . . . . . . . . . . . 84
Figure2.43 Time series for selected velocity channels of the 1665 MHz LCP com-
ponent in the intensive monitoring window. Features at 1.36 km/s, 1.45 km/s
and 1.53 km/s show a significant flare with a peak moderately variable in flux
density from 18 Jy to 28 Jy. The features in channels 3.64 km/s and 4.39 km/s
show no variation. Some data from 57820 MJD to 57890 MJD have been
flagged because of RFI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Figure2.44 Time series for selected velocity channels of the 1665 MHz RCP compo-
nent in the intensive monitoring window. The features in channels 1.31 km/s,
1.71 km/s, 1.88 km/s and 1.93 km/s show no variation in flux density. The
data from 57820 MJD to 57890 MJD have been flagged because of RFI. . . . . 86
Figure2.45 Time series for selected velocity channels of the 1667 MHz LCP compo-
nent in the intensive monitoring window. The features at 1.62 km/s, 1.67 km/s
and 1.75 km/s show significant variation of the peak from 10 Jy to 30 Jy whilst
the features at 2.19 km/s and 2.41 km/s show no variation. . . . . . . . . . . . 87
xv
 University of Ghana http://ugspace.ug.edu.gh
Figure2.46 Time series for selected velocity channels of the 1667 MHz RCP com-
ponent in the intensive monitoring window. The features in channels 1.4 km/s,
1.45 km/s, 2.15 km/s are only slightly variable with a slow change in ampli-
tude whilst the features in −0.84 km/s and −0.97 km/s are weak and show no
variation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Figure2.47 Time series for selected velocity channels of the 6668 MHz methanol
maser transition in the intensive monitoring window. The features in the veloc-
ity channels 1.18 km/s, 1.23 km/s, 1.27 km/s, 1.31 km/s and 1.36 km/s show
variability with a flare between 57780 MJD and 57860 MJD. . . . . . . . . . . 89
Figure2.48 Some selected channels of the 1665 MHz and 1667 MHz hydroxyl maser
lines in both polarizations together with the 6668 MHz methanol maser line
during the regular monitoring window. . . . . . . . . . . . . . . . . . . . . . . 91
Figure2.49 Some selected channels of the 1665 MHz and 1667 MHz hydroxyl lines
in both polarizations and the 6668 MHz methanol line during the intensive mon-
itoring window. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Figure3.1 The e-MERLIN interferometric array with seven antennas distributed
across the UK, with baselines spanning 11 km to 217 km. Displayed here
are the seven antennas, namely; Lovell, Mark 2, Cambridge, Defford, Knockin,
Darnhall and Pickmere (Belles et al., 2015). . . . . . . . . . . . . . . . . . . . 98
Figure3.2 Basic telescope arrangement in an interferometer. Adopted from Burke
and Graham-Smith (2002). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Figure3.3 Interferometery coordinates based on Burke and Graham-Smith (2002). . 100
Figure3.4 The uv coverage of e-MERLIN for the target during the observation. . . . 103
Figure3.5 Figure showing the elevation of the individual sources in the observa-
tion. Scans of the calibrators are coded by colour. The first field, coloured
green, is the flux calibrator followed by the bandpass calibrator coloured black.
The phase calibrator and the target are coloured brown and orange respectively.
And the last calibrator coloured pink is the polarisation calibrator. The phase
calibrator and the target have low elevations because of their positions on the sky.108
xvi
 University of Ghana http://ugspace.ug.edu.gh
Figure3.6 Plot of the delays for the entire duration of an observation on all the cal-
ibrators. Each antenna’s delay relative to the reference antenna is plotted in a
separate pane with the blue and green colour codes showing the different po-
larization components. Here Antid=0 represents Darnhall , Antid=1 represents
Knockin, Antid=2 represents Defford, Antid=3 represents Mark2, Antid=4 rep-
resents Pickmere, and Antid=5 represents Cambridge. . . . . . . . . . . . . . . 110
Figure3.7 Initial gain (phase) calibration solutions for all calibrators coloured by
spectral windows. Here Antid=0 represents Darnhall, Antid=1 represents Knockin,
Antid=2 represents Defford, Antid=3 represents Mark2, Antid=4 represents
Pickmere , and Antid=5 represents Cambridge. . . . . . . . . . . . . . . . . . 112
Figure3.8 Initial amplitude calibration solution for all calibrators. In the mosaic,
the first field represents the polarisation solutions, the second field represents
the flux calibrator solutions, the third field represents the bandpass solutions,
and the last field represents the scans of the phase calibrator solutions. Here
Antid=0 represents Darnhall, Antid=1 represents Knockin, Antid=2 represents
Defford, Antid=3 represents Mark2, Antid=4 represents Pickmere, and Antid=5
represents Cambridge. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Figure3.9 Flux density scale for the calibrators at L-band observing frequencies.
The blue line shows the polarisation calibrator, the green line shows the phase
calibrator and the red line shows the bandpass calibrator. The upper panel
shows the flux density and the spectral index of each calibrator at the observing
frequency. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Figure3.10 Amplitude gain solutions for all the calibrators, the scaled amplitudes
for each antenna are plotted in a separate pane. Each of the spectral windows is
plotted with two polarization components shown by the different colours. . . . 116
Figure3.11 Gain phase solutions of all the calibrators. Each of the spectral windows
for each calibrator is plotted with two polarization components shown by the
different colours. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
xvii
 University of Ghana http://ugspace.ug.edu.gh
Figure3.12 Phase solutions of the bandpass calibration on the bandpass calibrator
(1407+ 284). Each of the 8 spectral windows is plotted with two polarization
components shown by the different colours. . . . . . . . . . . . . . . . . . . . 119
Figure3.13 Amplitude solutions of the bandpass calibration on the bandpass calibra-
tor (1407+284). Each of the 8 spectral windows is plotted with two polariza-
tion components shown by different colours. The edges of the bandpass are
the fall off due to the instrumental demarcation of the frequencies by spectral
windows. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Figure3.14 Solutions of phase scan averages on the phase calibrator transferred onto
the target. Each colour represent a spectral window and a polarization compo-
nent averaged over a scan. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Figure3.15 Solutions of amplitude scan averages on the phase calibrator transferred
onto the target. Each colour represent a spectral window and a polarization
component averaged over a scan. . . . . . . . . . . . . . . . . . . . . . . . . . 122
Figure3.16 Image of the phase calibrator with a deconvolved size of (622.61±2.31) mas×
(129.9±0.10) mas at a position angle of (13.27±0.01) degrees. Shown in the
lower left corner of the image is the synthesized beam. . . . . . . . . . . . . . 123
Figure3.17 An image of G9.745+0.106 in the field of view which was used for the
self calibration with a deconvolved size of (456± 54) mas× (280± 17) mas
at a position angle of (67±12) degrees. Shown in the lower left corner of the
image is the synthesized beam of size 0.57 arcsecs×0.10 arcsecs at a position
angle of 11.49 degrees. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Figure3.18 Phase only solutions on G9.745+0.106 applied to the continuum with a
spell of bad weather recorded between 20 : 20 hours and 21 : 30 hours. Some
solutions were lost on Cambridge. . . . . . . . . . . . . . . . . . . . . . . . . 126
Figure3.19 Amplitude self calibration on G9.745+0.106 applied to the continuum
with a spell of bad weather recorded between 20 : 20 hours and 21 : 30 hours. . 127
xviii
 University of Ghana http://ugspace.ug.edu.gh
Figure3.20 Phase offset solutions between the narrow and wide band channelisation
at 1665 MHz and 1667 MHz. The two polarisation components are shown by
the different colours. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Figure3.21 Bandpass calibration on the narrow band data for the 1665 MHz maser
line. The calculated solutions are coloured by the two polarizations. . . . . . . 130
Figure3.22 Bandpass calibration on the narrow band data for the 1667 MHz maser
line. The calculated solutions are coloured by the two polarizations. . . . . . . 131
Figure3.23 Image of the field of the L-band radio continuum of G9.62+0.20E. The
pink cross shows the expected position of the radio continuum. Shown at the
lower left corner of the image is the synthesized beam of size 0.78 arcsecs×
0.57 arcsecs at a position angle of −25.36 degrees. . . . . . . . . . . . . . . . 133
Figure3.24 Spectral index plot of the G9.62 + 0.20E HII region. Shown in blue
squares are the flux density values from different high frequencies adapted from
Franco et al. (2000). The red line is the least squares fit to the higher frequency
detections only. Plotted in green circle is the 3σ upper limit on the flux density
for the G9.62+0.20E star forming region at 1.5 GHz. . . . . . . . . . . . . . . 134
Figure3.25 Image of the L-band radio continuum emission of G9.62+0.20C shown
in colour scale. Shown at the lower left corner of the image is the synthesized
beam of size 0.78 arcsecs×0.57 arcsecs at a position angle of −25.36 degrees. 135
Figure3.26 Image of the L-band radio continuum emission of G9.62+ 0.20D with
a deconvolved size of (2.090±0.106) arcsecs× (0.781±0.025) arcsecs and a
position angle of (1.9± 1.4) degrees shown in colour scale. At the lower left
corner of the image is the synthesized beam of size 0.78 arcsecs×0.57 arcsecs
at a position angle of −25.36 degrees. . . . . . . . . . . . . . . . . . . . . . . 136
Figure3.27 The flux density distribution for the 1.665 GHz LCP component over the
intensity range 7.5 Jy/beam to 100 Jy/beam. At the lower left corner of the
image is the synthesized beam. . . . . . . . . . . . . . . . . . . . . . . . . . . 138
xix
 University of Ghana http://ugspace.ug.edu.gh
Figure3.28 First moment map for the LCP component of the 1.665 GHz maser line
shown in colour scale. The maser spots in the 1.64 km/s and 1.95 km/s velocity
channels are displayed. At the lower left corner of the image is the synthesized
beam. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Figure3.29 The spectrum for the LCP component of the 1665 MHz OH maser line. . 139
Figure3.30 The flux density distribution for the RCP component of the 1.665 GHz
maser line over the intensity range 2.5 Jy/beam to 100 Jy/beam. At the lower
left corner of the image is the synthesized beam. . . . . . . . . . . . . . . . . . 140
Figure3.31 First moment map for the RCP component of the 1.665 GHz maser line
shown in colour scale. The maser spots in the 1.38 km/s velocity channel is
displayed. At the lower left corner of the image is the synthesized beam. . . . . 141
Figure3.32 The spectrum for the RCP component of the 1665 MHz OH maser line
showing the maser lines at the velocity channels 1.38 km/s, 2.12 km/s and
5.29 km/s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Figure3.33 The flux density distribution for the LCP component of the 1.667 GHz
maser line over the intensity range 4.5 Jy/beam to 100 Jy. At the lower left
corner of the image is the synthesized beam. . . . . . . . . . . . . . . . . . . . 142
Figure3.34 First moment map for the LCP component of the 1.667 GHz maser line
shown in colour scale. The maser spot in the 1.77 km/s velocity channel is
displayed. At the lower left corner of the image is the synthesized beam. . . . . 143
Figure3.35 The spectrum for the LCP component of the 1667 MHz OH maser line
showing the spectral features at −0.95 km/s, 1.77 km/s and 2.56 km/s. . . . . 143
Figure3.36 The flux density distribution for the RCP component of the 1.667 GHz
maser line over the intensity range 1.3 Jy/beam to 100 Jy/beam. At the lower
left corner of the image is the synthesized beam. . . . . . . . . . . . . . . . . . 144
Figure3.37 First moment map for the RCP component of the 1.667 GHz maser line
shown in colour scale. The maser spots in the 0.63 km/s, 1.69 km/s and
2.21 km/s velocity channels are displayed. At the lower left corner of the
image is the synthesized beam. . . . . . . . . . . . . . . . . . . . . . . . . . . 145
xx
 University of Ghana http://ugspace.ug.edu.gh
Figure3.38 The spectrum for the RCP component of the 1667 MHz OH maser line
showing the spectral features at velocity channels −0.46 km/s, 0.63 km/s,
1.69 km/s and 2.21 km/s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Figure4.3 Image of the phase calibrator with a deconvolved size of (234.59±0.30) mas×
(27.391±0.004) mas at a position angle of (7.721±0.001) degrees. Shown in
the lower left corner of the image is the synthesized beam. . . . . . . . . . . . 155
Figure4.1 Phase calibration solutions on the phase calibrator for day 20170717.
Each scan consists of the spectral windows plotted with the two polarization
components shown by the different colours. . . . . . . . . . . . . . . . . . . . 156
Figure4.2 Amplitude calibration solutions on the phase calibrator for day 20170717.
Each scan consists of the spectral windows plotted with two polarization com-
ponents shown by the different colours. . . . . . . . . . . . . . . . . . . . . . 157
Figure4.4 Phase component of the bandpass calibration on the narrow band data
coloured by the spectral windows and the two polarization components. . . . . 159
Figure4.5 Amplitude component of the bandpass calibration on the narrow band
data coloured by the spectral windows and the two polarization components. . . 160
Figure4.6 Phase self calibration solutions on the narrow band data coloured by the
two polarizations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
Figure4.7 Amplitude self calibration solutions on the narrow band data coloured by
the two polarizations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Figure4.8 Image of the C-band radio continuum of G9.62+0.20E shown in colour
scale. At the lower left corner of the image is the synthesized beam. . . . . . . 164
Figure4.9 Spectral index plot of the G9.62 + 0.20E HII region. Shown in blue
squares are the flux density values from different frequencies adapted from
Franco et al. (2000). The flux density for the C-band continuum is shown with
a green circle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
xxi
 University of Ghana http://ugspace.ug.edu.gh
Figure4.10 An updated spectral index plot of the flux density distribution of the
G9.62+0.20E region at different frequencies adapted from Franco et al. (2000).
The update to the plot is the flux density for the C-band continuum shown with
a yellow circle and the 3σ upper limit of the peak flux density at L-band also
shown with a green circle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Figure4.11 C-band radio continuum image of G9.62+0.20D shown in colour scale.
At the lower left corner of the image is the synthesized beam. . . . . . . . . . . 169
Figure4.12 The flux density distribution for the 6.7 GHz methanol masers over the
intensity range 700 Jy/beam to 1500 Jy/beam. At the lower left corner of the
image is the synthesized beam. . . . . . . . . . . . . . . . . . . . . . . . . . . 170
Figure4.13 First moment map for the 6.7 GHz methanol masers shown in colour
scale. The maser spot in the 1.25 km/s velocity channel is displayed. At the
lower left corner of the image is the synthesized beam . . . . . . . . . . . . . . 171
Figure4.14 The spectrum for the methanol masers at 6.7 GHz. . . . . . . . . . . . . 172
Figure4.15 The C-band radio continuum of the G9.62+0.20E with the positions of
the maser spots overlayed on it. The peak position of the 1.25 km/s maser
spot is represented by a red circle and the −0.26 km/s maser spot is marked
by a yellow star. The peak position of the 1.05 km/s maser spot in Sanna et al.
(2015) is shown with a black plus sign. . . . . . . . . . . . . . . . . . . . . . . 174
xxii
 University of Ghana http://ugspace.ug.edu.gh
LIST OF TABLES
Table1.1 Properties of molecular clouds (Shu et al., 1987). . . . . . . . . . . . . . 3
Table1.2 Properties of giant molecular clouds and the structures found within them
(Zinnecker and Yorke, 2007; Stahler and Palla, 2004). . . . . . . . . . . . . . . 4
Table2.1 Specification for the 26 m HartRAO telescope. . . . . . . . . . . . . . . . 38
Table2.2 Observing parameters for G9.62+0.20E at 1665 MHz and 1667 MHz. . . 41
Table2.3 Observing parameters for G9.62+0.20E at 6668 MHz. . . . 41
Table2.4 Point Source Sensitivity values calculated for LCP and RCP. . . . . . . . 55
Table2.5 Average flux density and rms noise for the regular monitoring window. . . 56
Table2.6 Average flux density and rms noise for the intensive monitoring window. . 57
Table2.7 Detected OH maser velocity channels and their flux densities for the reg-
ular monitoring window. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Table2.8 Detected OH maser velocity channels and their flux densities for the in-
tensive monitoring window. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Table2.9 Detected OH maser velocity channels and their flux densities derived from
Caswell et al. (2013). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Table2.10 Detected 6668 MHz maser velocity channels and their flux densities for
both regular and intensive monitoring windows. . . . . . . . . . . . . . . . . . 68
Table2.11 Detected 6668 MHz maser velocity channels and their flux densities de-
rived from Goedhart et al. (2004). . . . . . . . . . . . . . . . . . . . . . . . . 68
Table3.1 Data observation schedule and total on source time. . . . . . . . . . . . . 105
xxiii
 University of Ghana http://ugspace.ug.edu.gh
Table3.2 Observing parameters for the L-band continuum at a central frequency of
1.5 GHz. Antenna labels: De: Defford, Cm: Cambridge, Da: Darnhall, Kn:
Knockin, Mk2: Mark 2, Pi: Pickmere. . . . . . . . . . . . . . . . . . . . . . . 105
Table3.3 Observing parameters for the 1.6 GHz OH maser lines observation. . . . . 106
Table3.4 Calibrators used for the 1.6 GHz OH observations and their flux densities. 106
Table3.5 Position and flux density of phase calibrator. . . . . . . . . . . . . . . . . 123
Table3.6 Position and flux density of G9.745+0.106. . . . . . . . . . . . . . . . . 124
Table3.7 Positions and flux densities of the L-band continuum associated with source
C, D and E in G9.62+0.20. The expected position of the source E is adopted
from Sanna et al. (2015). For G9.62+0.20E the peak flux is given by the non-
detection upper limit. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Table3.8 Absolute positions, velocities and flux densities of the polarisations of the
mainline OH masers in G9.62+0.20E. . . . . . . . . . . . . . . . . . . . . . . 147
Table4.1 Data observation schedule and total on source time. . . . . . . . . . . . . 152
Table4.2 Observing parameters for the C-band continuum at a central frequency of
6.5 GHz. Antenna labels: De: Defford, Cm: Cambridge, Da: Darnhall, Kn:
Knockin, Mk2: Mark 2, Pi: Pickmere. . . . . . . . . . . . . . . . . . . . . . . 152
Table4.3 Observing parameters for methanol lines at 6.7 GHz. . . . . . . . . . . . 153
Table4.4 Calibrators used for the observations at 6.7 GHz and their flux densities. . 153
Table4.5 Position and flux density of Phase calibrator. . . . . . . . . . . . . . . . . 155
Table4.6 Position and flux density of the C-band continuum of G9.62+0.20E and
G9.62+0.20D. Also given are the beam dimensions and the position angles of
the maser spots. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Table4.7 Physical properties of G9.62+0.20E. The size of the Q-band continuum
comes from Sanna (2019) based on the results in (Sanna et al., 2015). . . . . . 167
Table4.8 Absolute positions, velocities and flux densities of the 6.7 GHz methanol
maser spots in G9.62+0.20E. . . . . . . . . . . . . . . . . . . . . . . . . . . 173
xxiv
 University of Ghana http://ugspace.ug.edu.gh
LIST OF SYMBOLS
L Solar luminosity
M Solar mass
Mcrit Critical mass for gravitational collapse
MJ Jeans mass
xxv
 University of Ghana http://ugspace.ug.edu.gh
LIST OF ABBREVIATIONS
ATCA Australian Telescope Compact Array.
CMC Cold Molecular Cloud.
CWB Colliding Wind Binary.
Dec Declination.
e-MERLIN Enhanced-Multi Element Radio Linked Interfer-
ometer Network.
FIR Far Infrared.
GMC Giant Molecular Cloud.
HartRAO Hartebeesthoek Radio Astronomy Observatory.
HMC Hot Molecular Cloud.
HR Hertzsprung-Russell.
IR Infrared.
ISM Interstellar Medium.
JVLA Jansky Very Large Array.
xxvi
 University of Ghana http://ugspace.ug.edu.gh
KH Kelvin-Helmholtz.
LBA Long Baseline Array.
LCP Left Circular Polarisation.
MC Molecular Cloud.
MIR Mid Infrared.
MS Main Sequence.
MYSO Massive Young Stellar Object.
PAH Polycyclic Aromatic Hydrocarbon.
PDR Photo-Dissociation Regions.
PMS Pre Main Sequence.
RA Right Ascension.
RCP Right Circular Polarisation.
RFI Radio Frequency Interference.
submm Submillimeter.
UCHII Ultracompact.
VLBA Very Long Baseline Array.
VLBI Very Long Baseline Interferometry.
ZAMS Zero Age Main Sequence.
xxvii
 University of Ghana http://ugspace.ug.edu.gh
CHAPTER 1
INTRODUCTION
High mass star formation involves the processes by which gases in the interstellar medium
(ISM) are transformed into stars of various masses greater than 8 M and luminosities greater
than 103 L (Urquhart et al., 2015; Guzman et al., 2016) either in isolation or in clusters.
During these processes matter and energy are continuously exchanged within the ISM, form-
ing a recycle hub of energy and matter. Hence, the formation of high mass stars is of great
importance for understanding the evolution of galaxies, and the quest to understand the physics
governing these processes has inspired a lot of research in the field of astrophysics.
The theory on how low mass stars (mass less than 8 M) form has greatly been shaped by
their theoretical and observational studies in our Galaxy (Shu et al., 1987). This has been possi-
ble because of the proximity of some low mass stars to Earth and their abundance in our Galaxy.
The closest low mass star forming region, Ophiuchus is located at about 120 pc (Loinard et al.,
2008). Our observation of how high mass stars form, however, is greatly hindered by vari-
ous factors, which are also part of the environmental conditions necessary for their formation.
First, high mass stars are rare, and availability of exemplars on which to base theoretical and
observational studies are severely limited. They are formed at large distances away from the
Earth, with the closest high mass star forming region being Orion KL at 437 pc (Genzel et al.,
1981; Hirota et al., 2007) and the next in proximity being Cepheus A at ∼ 700 pc (Moscadelli
and Reid, 2009; Dzib et al., 2011). The whole process of the formation of high mass stars is
shrouded in dense cocoons of gas and dust resulting in high extinction which makes the region
opaque to optical probes (Zinnecker and Yorke, 2007). Hence, their study relies heavily on ob-
servations of thermal dust emission at infra-red (IR) and submillimeter (submm) wavelengths,
1
 University of Ghana http://ugspace.ug.edu.gh
ionized hydrogen emission at radio wavelengths and spectral line emissions from the surround-
ing molecular gas clouds. Figure 1.1 is a three-colour composite image of the G9.62+ 0.20
high mass star forming region showing emissions at 8 µm in red, at 4.5 µm in green and at
3.6 µm in blue (Liu et al., 2017). Labelled A, B, C, D and E on the figure are the radio sources
as named by Garay et al. (1993), with the white contours being 450 µm emissions which trace
dust emissions from the region and the red contours being 8 µm emissions showing polycyclic
aromatic hydrocarbon (PAH) emissions tracing photo-dissociation regions (PDR’s) which hith-
erto have not been revealed with optical probes. In addition to all the factors given so far, the
relatively short time scales of important stages of high mass star formation present a problem
when tracking their evolutionary phases. For example, the Kelvin-Helmholtz (KH) timescale
(time it takes for a young stellar object or stellar core to radiate its thermal energy content) is
typically 107 yr for low mass stars but less than 104 yr for high-mass stars. To compound the
problem high mass stars are mostly found in clusters with their low mass counterparts, making
it difficult to observe them individually (Andrews et al., 2014).
Figure 1.1: A three-colour composite Spitzer image of the G9.62+ 0.20 complex showing
emissions at 8 µm in red, 4.5 µm in green, and 3.6 µm in blue. The contours are constant
intensity lines for which the white contours trace dust emissions at 450 µm and the red contours
trace photo-dissociation regions at 8 µm (Liu et al., 2017).
2
 University of Ghana http://ugspace.ug.edu.gh
In spite of the challenges in their observation, the critical role played by high mass stars in
the evolution of galaxies and the universe cannot be over emphasised (Hoyle, 1954). High mass
stars disperse large amounts of radiative energy across galaxies. They also serve as the origin
of elements, that is, they are the chemical factories of the ISM, producing atoms, molecules and
ionised gases, and thus are the principal source of heavy elements in the universe (Hoyle, 1954).
Despite decades of work a general theory of star formation that explains the birth of stars of all
masses emanating from different environmental conditions is still sketchy. Depending on their
masses, luminosities, and how they die, star formation is grouped into low mass star formation
for stars with M < 8 M that end their lives as white dwarves and high mass star formation for
stars with M > 8 M that end their lives in a Type II supernova.
Generally, star formation occurs in warm dense cores of giant molecular clouds (GMC)
which consist of cold dense molecular gas and dust concentrated in the spiral arms of a galaxy.
Molecular clouds (MC’s) are observed to be highly turbulent (McKee and Tan, 2003) with
properties as given in Table 1.1.
Table 1.1: Properties of molecular clouds (Shu et al., 1987).
Property Value
Mass range (M) 10 - 106
Size range (pc) 1 - 100
Density range (cm−3) 103 - 104
Temperature (K) 10 - 20
Age (yr) 105 - 109
Within the molecular clouds there are filamentary and hierarchical structures such as clumps,
cold molecular cores (CMC) and hot molecular cores (HMC) (Arzoumanian et al., 2011; Hacar
et al., 2013; Smith et al., 2014; Wang et al., 2019). These structures are made up of different
density regions with typical densities, sizes (McGuire et al., 2016) and other important proper-
ties as given in Table 1.2 (Zinnecker and Yorke, 2007; Stahler and Palla, 2004). These regions
of high density gradients of compressed gas and dust occur as a result of a complex interplay
between gravity and several opposing forces such as turbulence, gas pressure, magnetic fields
and radiative pressure (Traficante et al., 2020).
To form a star of average density 103 kg m−3 from a GMC density of order 10−18 kg m−3,
the GMC must be sufficiently compressed which can only be possible by self gravity; thus,
3
 University of Ghana http://ugspace.ug.edu.gh
Table 1.2: Properties of giant molecular clouds and the structures found within them (Zinnecker
and Yorke, 2007; Stahler and Palla, 2004).
Property GMC Clumps CMC HMC
Mass (M) 104 - 106.5 102 - 103 100 - 500 < 100
Size (pc) 10 - 100 1 - 1.5 0.2 - 0.3 0.1
Mean density (cm−3) 50 - 500 103 - 104 104 - 105 ≈ 107
Temperature (K) ≈ 10 - 20 10 - 30 8 - 12 > 100
gravity must overcome the turbulence, gas pressure, magnetic fields and radiative pressure.
Consider a spherical shell of radius r and mass dM(r) in a single isolated spherically symmetric
cloud. At equilibrium the outward pressure on the shell will be equal to the force of gravity
acting on the shell, giving
2 −GM(r)4πr dP(r) = 2 dM(r) (1.1)r
where G is the gravitation constant (Stahler and Palla, 2004). Integrating from centre to surface
over a sphere of volume V (Rc) = 4 33πRc gives
∫ P ∫s 0 GM(r)
3 V (r)dP(r) = dM(r) (1.2)
P0 Mc r
where Ps and P0 are the pressures at the surface and center, Mc is the cloud mass and Rc is the
cloud radius. The total gravitational energy content of the cloud therefore is
∫ Mc
Ω =− GM(r) dM(r) (1.3)
0 r
and for an ideal mono-atomic gas the mean kinetic energy is
3
U = Nk
2 B
T (1.4)
where kB is the Boltzmann constant, T is the uniform temperature of the cloud and N is the
total number of molecules in the cloud. For a cloud with constant density ρ , constant pressure
Pc and zero external pressure, Ps = 0 and U is the total thermal energy content of the cloud,
yielding the virial equation
4
 University of Ghana http://ugspace.ug.edu.gh
2U +Ω = 0 (1.5)
If 2U =−Ω, the cloud is stable and in equilibrium. When 2U > Ω, pressure overcomes grav-
ity and the cloud is dispersed. When 2U < −Ω, gravity overcomes pressure and the cloud
contracts. Thus, for contraction
3 GM2
> 3VcPc (1.6)5 Rc
where P = Nk T/V and R = (3M /4πρ )1/3c B c c c c have been used. Substituting N = Mc/µMH ,
where µMH is the mean molecular mass of the gas in the cloud in Equation 1.6 leads to a
critical mass
( )3/2( )5kBT 3 1/2Mcrit = (1.7)GµMH 4πρc
above which the cloud collapses; that is, the√cloud collapses if Mc > Mcrit. The critical mass
for collapse is proportional to v3T , where vT = kBT/µMH is the isothermal speed of sound in
the cloud.
For a cloud undergoing collapse (ignoring pressure and assuming gravity dominates) the
equation of motion for a thin shell with an initial radius Rc is
d2r −GM(r)
3
−4πGRc ρc2 = 2 = 2 (1.8)dt r 3r
Assuming the mass and acceleration remain constant, a free-fall time of
( )
3π 1/2
t f f = (1.9)32Gρc
is obtained. The free-fall time is long for low-density regions and short for high-density regions.
Thus the virial equation, Equation 1.5, sets the stability criterion of a molecular cloud supported
by thermal pressure alone. A cloud supported only by thermal pressure will collapse if its mass
exceeds the Jeans mass
5
 University of Ghana http://ugspace.ug.edu.gh
MJ ∝ (T )3/2(ρ)−1/2 (1.10)
(See Equation 1.7.) Further contraction of a collapsing star results in approximately half of the
gravitational energy gained being used to increase the internal temperature of the cloud and
the residual energy being radiated into space. The duration for this further contraction is the
Kelvin-Helmholtz (KH) time given by
GM2
tKH = (1.11)L∗R∗
where L∗ is luminosity of the the star and R∗ is its radius. This duration determines how quickly
a proto-star contracts during its pre-main sequence (PMS) phase before it enters the zero age
main sequence (ZAMS) stage and hydrogen burning in its core through fusion reactions sub-
sequently begins. It then joins the main sequence (MS) stars on the Hertzsprung-Russell (HR)
diagram.
The paradigm of low mass star formation as given by Shu et al. (1987) describes the se-
quence of events for isolated cases of low mass star formation by linking the theories of low
mass star evolution to the observed situation reported by Adams et al. (1987). The scheme of
events was extended by Lada (1985) and Andre et al. (1993) to include much earlier states of
young stellar objects when they are still hidden in their natal clouds and can only be seen at
FIR and submm wavelengths as in Figure 1 of Feigelson and Montmerle (1999) and Figure 1.1
of Isella (2006). In this model, the formation of a low mass star is divided into four distinct
stages. The first stage involves slowly rotating cloud cores being formed from a molecular
cloud as shown in Figure 1.2(a). This occurs by ambipolar diffusion resulting from the slow
drift of ions and neutral atoms which causes the clouds to lose magnetic and turbulent support.
In addition, clumps of mass greater than the critical mass for collapse may also fragment as
they contract and flatten (Shu et al., 1987; Zhang et al., 2015).
The second stage sets off when the cloud core becomes unstable in a dynamic inside out
collapse as in Figure 1.2(b). The slowly rotating cloud of gas then contracts to form a central
proto-star. The proto-star then enters the third stage of evolution which is the bipolar outflow
6
 University of Ghana http://ugspace.ug.edu.gh
Figure 1.2: The four stages of the low mass star formation process as illustrated by Shu et al.
(1987); van Dishoeck and Blake (1998); Greene (2001). Plate (a) depicts the fragmentation of
a molecular cloud into pre-stellar cores as the cloud collapses under its gravity. In plate (b)
the proto-star builds up mass from in-falling gas as it continues to collapse. Plate (c) shows
the formation of a disc in the nebula as the proto-star conserves angular momentum and the
breakout out of bipolar outflows along the the axis of rotation of the T-Tauri star. Finally the
nebula is swept away in Plate (d) revealing the newly formed pre-main sequence star and the
remains of the accretion disk.
phase in Figure 1.2(c) (Swift and Welch, 2008; Wang et al., 2014). The cloud of gas contracts
further, forming a rotating sphere, and to conserve angular momentum, flattens into an accretion
disk surrounded by an in-falling envelope of dust and gas (Luhman et al., 1998). The proto-
star accretes incoming material which falls onto the disk rather than onto the center (Dunham
and Vorobyov, 2012). Fast moving gas molecules are driven along the rotational axis in the
form of collimated jets and bipolar outflows. As time progresses, the opening of the stellar
outflows sweeps across all directions revealing the fourth stage of a young stellar object (YSO)
surrounded by a remnant nebular disk as can be seen in Figure 1.2(d) (Chen et al., 2016). As
accretion stops the PMS star begins its journey of evolution on the HR diagram.
7
 University of Ghana http://ugspace.ug.edu.gh
High mass stars on the other hand do not follow the process described above in their forma-
tion. Since both low and high mass stars are formed in regions of similar densities they should
have the same free fall time (the timescale it takes for a pressure-free cloud to collapse dynam-
ically to a very small size under its own gravity) and acrshortKH time. For the relatively short
life span of high mass stars the theoretical considerations of their formation run into problems
when the free fall time is comparable to the KH time. The KH time is short for high mass stars
implying they arrive on the main sequence while still embedded in their molecular clouds and
still accreting matter and radiating before collapse is complete, in spite of their high luminosity
which creates an outward radiation pressure that should stop matter from being accreted on to
the new star (Beuther et al., 2007).
Theoretically three models have been advanced to explain the formation mechanisms of
massive stars. These are monolithic collapse and turbulent core accretion models, competitive
accretion and runaway growth models, and stellar collision and binary merger models. In the
monolithic collapse and core accretion models, the high turbulence of the cores enable them
drive a high accretion rate in order to build a massive star. This massive star formation model
remains a generalization of the low mass case, but with the mass of the collapsing core related
to the mass of the final star (Yorke and Sonnhalter, 2002; McKee and Tan, 2003; Krumholz
et al., 2010). This model does not address why the core of the massive proto-stellar object does
not further fragment into small stars but remains massive (Dobbs et al., 2005).
The competitive accretion and run away growth model invokes the cluster nature of high
mass star forming regions to form massive stars via accretion onto a central proto-star in the
proto-cluster by gravitational collapse in which the central proto-star has access to more matter
falling into the center (Zinnecker, 1982; Bonnell et al., 1997, 2004; Bonnell and Bate, 2006).
Here the central argument is that accretion is dependent on the large scale flows within the
gravitational potential of the molecular clumps. Stars located at the centre of the gravitational
potential are thus able to acrete more materials. This theory does not explain why accretion is
not halted in the face of such high radiative pressure.
In an attempt to overcome the problem of radiation pressure the stellar collision and binary
merger model suggests that massive stars form by direct collisions of lower mass stars. How-
8
 University of Ghana http://ugspace.ug.edu.gh
ever, the process requires high initial stellar densities (Bonnell and Davies, 1998; Bonnell et al.,
2003). This model has been discredited largely due to the extremely high cluster densities re-
quired and moreover this mechanism is not required to get over the issue of accretion given the
work that is later described on Rayleigh- Taylor instabilities (Kuiper et al., 2013).
These formation mechanisms largely depend on stellar formative environment, that is,
whether the stars form in isolation or in clusters. These mechanisms also seek to answer the
questions of how proto-stellar cores avoid further fragmentation, how massive proto-stars gain
their final mass in the face of great radiative pressure and subsequently how binary stars form
(Krumholz, 2014, 2015a). Simulations in 2D and 3D have been used to explore the effect of
radiative transfer and magnetic fields on star formation (Commerçon et al., 2011; Meyers and
Mckee, 2013). The results confirm the suppression of cores from further fragmentation. On
the issue of radiative pressure halting accretion, molecular gas outflows at high velocities in
opposite directions present mechanisms through which radiation can escape whilst the star still
accretes material via the disk.
The effects of disks (compact flattened structures around the luminous object) (Kuiper et al.,
2016; Maud et al., 2018), outflows (Kolligan and Kuiper, 2017) and jets on radiative pressure
(Kuiper et al., 2013) have been simulated. Observationally the existence of disks around mas-
sive stars have been revealed with radio investigations (Sanchez-Monge et al., 2014; Johnston
et al., 2015; Beuther et al., 2017), and the search for jets have also proved successful (Guzman
et al., 2016; Purser et al., 2016). Outflows have also been observed in some massive stellar ob-
jects (Beuther et al., 2007,b). The effect of radiation pressure is accounted for in the simulation
when in-falling materials strike the disk at great distances and are not heated at sublimation
temperatures but are shielded from the pressure so the radiation can escape along the perpen-
dicular outflow channels (Krumholz, 2015b). The disk efficiently dissipates the radiation by
absorbing the photons and re-radiating them from its surface (Tan et al., 2014; Stahler and
Palla, 2004).
Observationally, before the main sequence stage, the processes of massive star formation are
embedded in obscuring clouds (Traficante et al., 2015; Eden et al., 2019; Svoboda et al., 2019).
The stages of events, shown in Figure 1.3 (Purcell et al., 2006; Breen et al., 2010), involve
9
 University of Ghana http://ugspace.ug.edu.gh
IR dark molecular clouds (Figure 1.3(i)) of temperature of about 15 K and mass about 100 M
emitting strongly in the submm spectral range due to cold dust. These IR dark molecular clouds
may or may not posses starless cold cores (Sridharan et al., 2005; Urquhart et al., 2008; Peretto
and Fuller, 2009).
Figure 1.3: The observed evolutionary stages of high mass stars after the fragmentation and
collapse of molecular clouds (i). The prestellar cold collapsing cores (ii) develop into hot
molecular cores (iii), which are heated by the embedded massive star. An ultracompact HII
region (region of ionised gas) is formed around the young star (iv) which evolves into a diffuse
HII region in OB associations (v) (Purcell et al., 2006; Breen et al., 2010).
It is hypothesised that isothermal collapse of the cold cores as in Figure 1.3(ii) under
their own gravity increases the core density, releases gravitational potential energy as heat
and thereby increases the kinetic energy of the molecules which results in the formation of
hot molecular cores (see Figure 1.3(iii)). With a growing proto-star at the center and outflows
from its poles (Cesaroni et al., 2006; Pestalozzi et al., 2004) the hot molecular core begins to
warm up its surrounding dust and gas, resulting in the evaporation of molecules formed from
the warmed dust grains (Menten, 1991; Walsh et al., 1998; Viti and Williams, 1999; De Buizer
and Minier, 2005; Hill et al., 2005; Linz et al., 2005). The proto-star, now a massive young
stellar object (MYSO) still embedded in the cloud becomes bloated (swollen star model) as its
radius increases by rapid accretion at high rates and pulsates (Hosokawa and Omukai, 2009;
Hosokawa et al., 2010; Davies et al., 2011; Liu et al., 2011). The MYSO then disperses its
envelope along with the outflows and emits in the mid infra-red (MIR) and FIR. As accretion
ends whilst the MYSO contracts to the main sequence phase, hydrogen burning begins and
10
 University of Ghana http://ugspace.ug.edu.gh
the Lyman continuum produced (Hoare, 2002; Hoare et al., 2007, 2005) starts to ionise the
surrounding atomic and neutral hydrogen as seen in Figure 1.3(iv) leading to the formation of
ultra compact HII (UCHII) and classical HII regions (Kurtz, 2005; Hoare and Franco, 2007).
Usually, massive stars are formed in OB association as in Figure 1.3v (Billington et al., 2018).
The RMS survey (Urquhart et al., 2008, 2013) also points to a high mass classification scheme
as outlined by a review of the observed stages which follows the sequence of events in the
evolution of high mass stars (Zinnecker and Yorke, 2007).
1.1 HII regions in massive star formation
As massive stars evolve and start burning hydrogen in their core they produce copious amounts
of far ultraviolet (FUV) radiation (Morton, 1967; Keto, 2007) which is absorbed by gas and
dust. The ionised Ultracompact (UCHII) regions produced by the interactions of the UV ra-
diation with their surroundings uniquely trace massive star formation (Wood and Churchwell,
1989; Walsh et al., 1997, 1999). The HII regions emit strongly at radio wavelengths via free-
free emission.
1.2 HII regions
An HII region is a localised volume of gas surrounding a massive star in which all the hydrogen
has been ionised. As the star contracts and heats up its UV output increases. The UV photons
with energies above the Lyman limit are absorbed by the atomic hydrogen surrounding the
star in a process called photo-ionisation. Photo-ionisation is balanced by recombination which
occurs when ions capture free electrons. Thermal pressure causes the ionised region to expand.
If the density of the region is constant, the HII region forms a sphere (Strömgen sphere) of
ionised gas with radius rs (Strömgen radius) having a well defined boundary called an ionisation
front (Strömgren, 1939). At equilibrium, the Strömgen radius is given by
( )
3Q 1/3
r ∗s = (1.12)4παn2H
11
 University of Ghana http://ugspace.ug.edu.gh
where Q∗ is the number of ionising photons emitted per second, nH is the gas density and α
is the recombination rate coefficient which is weakly dependent on temperature (Strömgren,
1939).
1.2.1 Free-free emission
Thermal bremsstrahlung from ionized hydrogen, also called free–free emission, results from
interactions between charged particles in thermal equilibrium; in this case a free electron is
accelerated by a proton. The electrons are free before and after the interaction. The strength and
spectrum of radio emission from an HII region depends on the distributions of the velocities of
the electrons, their temperature and other collision impact parameters. The resulting emission
from free-free interactions has a broad spectrum across a range of frequencies.
In Figure 1.4, the shape of the spectrum for free-free emissions from a spherical, uniform
and isothermal HII region is shown. This follows a black body spectrum at low frequencies
when it is optically thick and opaque with optical depth τ  1. The spectral intensity at low
frequencies correspond to the optically thick region which varies with frequency quadratically
up to a turn over frequency νturnover where the optical depth is τ = 1 depicted by a spectral break
(Wilson et al., 2009; Condon and Ransom, 2016). At high frequencies, where the region is
optically thin and transparent, the emission varies approximately as ν−0.1 and the optical depth
τ  1 (Burke and Graham-Smith, 2002). The flux density at high frequencies is proportional
to the emission measure along the line of sight of the HII region and is given by
∫ L
EM = (ne)2 ds (1.13)
0
where ne is the electron density for the region. EM is a factor contained in the specific intensity
Sν . The optical depth is given by
( )
T −1.35( ( )− e ν )−2.1τ = 3.28×10 7 EMν 4 (1.14)10 K GHz pc cm−6
where Te is the electron temperature (Mezger and Henderson, 1967; Condon and Ransom,
2016). The electron temperature Te in addition to the optical depth τ gives the brightness
12
 University of Ghana http://ugspace.ug.edu.gh
Figure 1.4: The ideal radio spectrum of an HII region. The solid line shows the free-free
emission from the UCHII region and the dashed line depicts free-free emission from the HCHII
region. The vertical dotted line separates the optically thick part from the optically thin part of
the UCHII region (Kurtz, 2005).
temperature
Tb = T τe (1− e ) (1.15)
Hence, at low frequencies the brightness depends on the electron temperature whilst the
brightness depends on the EM at high frequencies (Wilson et al., 2009).
13
 University of Ghana http://ugspace.ug.edu.gh
1.2.2 Classification of HII regions
The expansion of the HII region caused by the thermal pressure depends on the density dis-
tribution of the region (Franco et al., 2000). HII regions have been classified into classical,
compact, ultracompact, hypercompact and extended based on their angular size and the posi-
tion of the ionisation front. The hypercompact HII regions are the extremely young objects
confined to very small sizes of about 0.003 pc. The ultracompact HII regions are the imme-
diate high pressure environment of the star within its parental molecular cloud and are about
0.1 pc in size (Kurtz et al., 1994, 2000). Compact HII regions are those carving their way out
of the clouds with low densities and are located in more evolved zones of the clouds with sizes
0.1 pc-0.3 pc. When they mature they have sizes of up to several parsecs and these are the
extended HII regions. The expansion of these objects is important for our understanding of the
structure and fate of clouds with active star formation and the objects themselves. HII regions
are often associated with maser emission and are surrounded by dense molecular gas.
1.2.3 Molecule formation in the interstellar medium
In the ISM a large number of chemical reactions are responsible for the formation of molecules.
Chemical bonds are formed by radiative association, grain surface formation and associative
detachment. Bonds are also broken by photo-dissociation, dissociative recombination and colli-
sion dissociation. There are also rearrangement reactions consisting of ion molecule exchange,
charge transfer and neutral collisions (Tennyson, 2003; Caselli and Ceccarelli, 2012). The
dense molecular clouds have temperatures as low as 10 K-100 K. They also contain matter in
the form of grains of dust made up of carbon rich grains or silicates. At low temperatures the
cores also grow mantles of other molecules forming ice mantles. The grain surfaces provide the
path by which atomic hydrogen combines to form other chemicals on grain surfaces (Hatchell
et al., 1998; Tielens, 2013). During the hot molecular core phase of massive star formation the
dust temperature becomes high enough to sublimate molecules from the ices, injecting them
into the gas phase. The molecules considered in this thesis are OH and methanol.
14
 University of Ghana http://ugspace.ug.edu.gh
1.2.3.1 Hydroxyl molecule formation pathways
The hydroxyl molecule (OH), shown in Figure 1.5, is formed by oxygen hydrogenation on
grain surfaces via the reaction (Willacy and Williams, 1993; Das et al., 2008)
Figure 1.5: Model of the hydroxyl molecule.
H+O→ OH (1.16)
Alternatively UV photons from massive stars are absorbed by the H2O molecules in the ice. On
absorption of a Lyman photon the H2O molecule dissociates into H and the OH radical which
is retained on the mantle:
H2O→ H++OH− (1.17)
The H atom carries most of the excess energy in the dissociation reaction in Equation 1.17. The
OH− radical releases an electron to form an OH molecule (Yamamoto, 2017; Viti et al., 2000):
OH−→ OH+ e− (1.18)
1.2.3.2 Methanol molecule pathways
In the ISM methanol molecules form by grain surface reactions (Das et al., 2008). One such
reaction is the evaporation of molecular ices on dust grains (Viti and Williams, 1999). CO
molecules adsorbed onto dust grains are subjected to grain-surface reactions which result in the
formation of CH3OH molecules in the grain mantles (Sakai et al., 2010):
15
 University of Ghana http://ugspace.ug.edu.gh
H+CO→ HCO (1.19)
H+HCO→ H2CO (1.20)
H+H2CO→ H3CO (1.21)
H+H3CO→ CH3OH (1.22)
Figure 1.6 shows the molecular structure of the methanol molecule. The thermal emission
lines of methanol are a good probe of the physical and chemical conditions in the star forming
regions (Cragg et al., 2005).
Figure 1.6: Model of the methanol molecule.
1.3 Astrophysical masers
Astrophysical masers are a coherent source of amplified microwave radiation from certain
molecules in the ISM. Masers are produced terrestrially by the amplification of microwave
radiation in cavities whereas they are produced naturally under the necessary conditions of
seed radiation, population inversion, sufficient amplification path length, and a low velocity
gradient along the amplification path (velocity coherence) in space. Masers are compact, have
high brightness temperature, are variable (Caswell, Vaile and Ellingsen, 1995), and can be po-
larised (Elitzur, 1998; Trung, 2009). Their polarisation can help measure the magnetic field
16
 University of Ghana http://ugspace.ug.edu.gh
strengths of their environment.
Maser emissions arising from molecular clouds have become important tools for the study
of the chemical components and physical processes of their environment (Minier et al., 2003;
Edris et al., 2007). Maser species such as water, hydroxyl, and methanol trace a number of
stellar phenomena such as shocks, outflows, infall and circumstellar disks as high-mass stars
advance in the early stages of their evolution. It has been suggested that masers are not asso-
ciated with different evolutionary stages of massive stars (De Buizer, 2003; De Buizer et al.,
2005). However, the role of masers in tracing different evolutionary stages is still debated
(Ellingen et al., 2011). Hydroxyl masers in massive star forming regions (SFRs) usually have
associated UCHII regions and often occur in close association with class II methanol and water
masers (Edris et al., 2007). They trace shocked neutral gas ahead of expanding UCHII regions
(Fish and Reid, 2006) and jets (Argon and Reid, 2003) and serve as indicators of the velocity
and magnetic fields on a very small scale (Fish and Reid, 2006).
Methanol masers are divided into two classes (class I and class II) depending on the strongest
maser transition (Batrla et al., 1987; Menten, 1991). Class I methanol masers are usually found
between molecular outflows (Menten, 1991) and the parent cloud apart from compact contin-
uum sources and trace an earlier stage than class II methanol masers (Ellingsen, 2006). They
are pumped by collisions between the outflowing materials and the ambient interstellar material
occurring at the interfaces of shock fronts. Class II methanol masers, which are associated with
the proto-star phase, trace hot molecular cores and are also found close to UCHII regions (Ba-
trla et al., 1987) though they sometimes occur earlier than the UCHII phase (Minier et al., 2005;
Ellingsen, 2006). Class II masers are pumped by the intense radiation of the warm dust heated
by the proto-star and manifest at frequencies of 6.7 GHz, 12.2 GHz, 19.9 GHz, 23.1 GHz,
28.9 GHz, 37.7 GHz, 38.2 GHz, 38.4 GHz, 107.0 GHz, 108.8 GHz, 156.6 GHz and 157 GHz
whereas class I methanol masers emit at 9.9 GHz, 25.0 GHz, 36.2 GHz, 44.1 GHz, 84.5 GHz,
95.2 GHz and 146.6 GHz.
17
 University of Ghana http://ugspace.ug.edu.gh
1.3.1 Maser theory
The principles governing maser emission include radiative transfer, population inversion, opti-
cal depth, saturation and pumping which are explained below.
1.3.1.1 Radiative transfer
A travelling radio wave of intensity Iν encounters a cloud of matter (neutral gas) and interacts
with it. The cloud of matter may absorb some of the radiation, scatter some, and further emit
some of the radiation. These processes change the intensity of the transmitted radiation (Elitzur,
1992). In the case of pure absorption, a beam of area dA propagating a distance ds into matter
of absorption cross-section σν and number density n suffers a change in intensity
dIν =−αν Iν ds (1.23)
where αν = nσν is opacity. The transmitted intensity after propagating a finite distance in the
cloud is then
I −τν ∝ e ν (1.24)
where τν is the optical depth for a perfectly absorbing medium. The attenuation suffered by the
beam is termed optical depth τν and is defined by
∫
τν = αν ds (1.25)
τν is a measure of the absorption as the intensity of the beam decreases through the cloud. The
medium is optically thin when chances of interaction of the photons in the beam with particles
of the medium are small. For this the value of τν is less than unity. The medium however is
optically thick if the photons in the beam have a high chance of interacting with the particles of
the medium. Here τν is greater than unity and no radiation passing through that region reaches
the observer.
On the other hand an atom after absorbing a photon and going into a higher energy level
18
 University of Ghana http://ugspace.ug.edu.gh
Figure 1.7: An excited atom de-exciting from a high energy level to a low energy level via the
spontaneous emission of a photon.
(excited state) may de-excite to a lower energy level by spontaneously emitting a photon with
energy equal to the difference in energy of the levels as shown in Figure 1.7. In another mode
of de-excitation as shown in Figure 1.8, a photon of the same wavelength called the seed photon
interacts with an atom already in its excited state. The photon stimulates the atom into emitting
a photon to return to its lower energy state. The new photon’s phase, frequency, polarization,
as well as direction of travel are identical to those of the photon that stimulated the emission.
The photon emitted by the stimulated emission process can further stimulate other atoms in a
progressive cascade of emission, thereby amplifying the intensity of the incident radiation.
Figure 1.8: An incoming photon of a specific frequency interacts with an excited atom, causing
it to de-excite to a lower energy level. The emitted photon is identical to the incident photon.
The process of stimulated emission is akin to negative absorption and leads to growth of
beam intensity when the net gain it provides exceeds absorption. Thus, stimulated emission
accounts for amplification of radiation as it propagates through the cloud. The emission coeffi-
cient, εν , or emissivity, is defined as the energy emitted per unit time per unit volume per unit
19
 University of Ghana http://ugspace.ug.edu.gh
solid angle per unit frequency; hence, on crossing the path length ds, the intensity of the beam
is increased by
dIν = εν ds (1.26)
Combining the processes of emission and absorption defines the radiative transfer equation
dIν
=−αν I + εν (1.27)ds
which describes how the intensity of radiation varies as it propagates in a medium.
1.3.1.2 Population inversion
Figure 1.9 shows a two-level system under normal conditions of thermal equilibrium. The
number of particles in an energy level E is proportional to the Boltzmann factor e−E/kBT for
that level where T is the equilibrium temperature.
Figure 1.9: The different processes that populate or depopulate the levels in a collection of two
level systems in thermodynamic equilibrium. The radiative processes are absorption character-
ized by coefficient B12, spontaneous emission characterized by coefficient A21 and stimulated
emission characterized by coefficient B21. Collisional processes are characterized by coeffi-
cients C12 for excitation and C21 for de-excitation.
20
 University of Ghana http://ugspace.ug.edu.gh
In thermal equilibrium, the number, n2, of particles in the higher level E2 is always less than
the number, n1, of particles in the lower level E1. Hence, the population ratio, given by
n2 g2 −hν
= e kT < 1 (1.28)
n1 g1
satisfies
n1 n2
> (1.29)
g1 g2
where the gi’s are the statistical weights of the upper and lower states. The emissivity for spon-
taneous emission assumes the same line profile function φ(ν) which governs the absorption of
a photon and is given by
hν
εν = n2A21 φ(ν) (1.30)4π
where A21 is the Einstein coefficient for spontaneous emission. The absorption coefficient is
given by
hν
αν = (n1B12−n2B21) φ(ν) (1.31)4π
where B12 is the Einstein coefficient for absorption and B21 is the Einstein coefficient for stim-
ulated emission. The Einstein coefficients are related by the equations
A 2hν321
= 2 (1.32)B21 c
and
g
B 2
B21
12 = (1.33)g1
where h is the Planck constant and c is the speed of light in vacuum. In order for the net emis-
sion to be stimulated the rate of stimulated emission must be greater than the rate of absorption
which is not achievable in thermal equilibrium. Population inversion only occurs if the rate
at which particles move from the n2 population is less than the rate at which particles move
21
 University of Ghana http://ugspace.ug.edu.gh
to n1 implying that n2 should always be greater than n1. A mechanism that removes a suffi-
cient number of particles from the lower to the higher energy level thereby inverting the normal
population is needed which will lead to
n1 n2
< (1.34)
g1 g2
resulting in a net gain. Thus the absorption coefficient becomes negative and the gas cloud adds
to the radiation field and hence amplification of the radiation occurs.
1.3.1.3 Pumping
Population inversion in the energy levels of molecules is a necessary condition for maser action.
The population inversion from radiative, collisional or chemical pumping may result from small
departures of the normal population from the Boltzmann distribution. This occurs naturally in
the ISM due to low gas densities, thus, a combination of dense molecular gas and dust in
addition to either strong IR radiation fields (source of excitation) or long collisional times in
outflows or both leads to pumping of molecules to the desired higher levels in certain molecules
such as OH, H2O, and CH3OH (Elitzur, 1976, 1992). At low densities purely radiative pumping
occurs whereas collisional excitation results in purely collisional pumping. A combination of
both radiative and collisional pumping occurs in some molecules.
For any molecule, the total angular momentum J without the nuclear spin arises from cou-
pling of the electronic orbital angular momentum L, end-over-end rotation K and the electronic
spin S quantum numbers. The OH molecule in Figure 1.5 with an unpaired electron occupying
the 2pπ orbital for instance has a total angular momentum quantum number (exclusive of nu-
clear spin) given by J = 1±1/2 in the rotational ground state (K = 0) which leads to the two
2π states designated by 2π3/2 and 2π1/2.
Figure 1.10 shows the two allowed rotation ladders associated with the 2π 23/2 and the π1/2
rotational ground states of OH. Each level in the ladder, designated by its J value, is split by
interaction of the electronic motion and the molecular rotation into doublets (lambda doubling)
labelled ’+’ or ’−’ on the Figure. The lambda doublets are further split by the nuclear hyperfine
interaction to give the energy levels shown in Figure 1.10 between which the OH microwave
22
 University of Ghana http://ugspace.ug.edu.gh
Figure 1.10: Energy levels of the two ladders of the rotational states of OH. The splitting of the
ground state of OH occurs via lambda doubling and hyperfine interactions. The broken lines
show the pumping scheme for the 1612 MHz maser transition in OH (Elitzur, 1976).
transitions of interest occur (Elitzur, 1976). Figure 1.11 shows the hyperfine levels of the
2π3/2,J = 3/2 state. Each level is labelled with its total angular momentum quantum number F
which includes the nuclear contribution. Allowed transitions between these hyperfine levels are
governed by the selection rules4J = 0 and4F = 0,±1 which give rise to the four components
of the hydroxyl emission at 18 cm as shown in Figure 1.11
23
 University of Ghana http://ugspace.ug.edu.gh
Figure 1.11: Part of the rotational energy level diagram of the hydroxyl ground state showing
the hyperfine splitting of the rotational states that give rise to the 18 cm hydroxyl emission.
Shown are the four hyperfine masing transitions of the OH ground state.
The transitions for which 4F = 0 give rise to the 1665 MHz and 1665 MHz maser lines
collectively called the mainlines. The 1612 MHz and 1720 MHz transitions occur when4F =
±1 and are called the satellite lines.
For the 1612 MHz OH line, warm dust radiating at a wavelength of 35 µm excites the
OH molecule from its ground state 2π3/2,J = 3/2 manifold to the excited state 2π1/2,J = 5/2
manifold as shown in Figure 1.10. A metastable state is populated via the 2π1/2 ladder. The final
2π ,J = 1/2→ 21/2 π3/2J = 3/2 transition causes the inversion of the ground state population.
The methanol model in Figure 1.6 shows that the bond between the methyl and the hydroxyl
groups of the methanol molecule constitutes an axis about which the methyl group rotates
relative to the OH group. This rotation is constrained by interaction between the H atoms of
the two groups, with the result that the rotation is reduced to torsional oscillations about the
O–H bond. The coupling of the angular momentum associated with the torsional motion to the
total angular momentum results in two torsional symmetry states which lead to the A-type and
E-type methanol maser transitions as shown in Figure 1.12.
24
 University of Ghana http://ugspace.ug.edu.gh
Figure 1.12: Energy level diagram for E-type (left) and A-type (right) symmetry states of the
ground state methanol transitions. The E-type symmetry for which k < 0 indicates transitions
for CH −3OH E levels whereas the E-type symmetry for which k > 0 indicates transitions for
CH +3OH E levels. The A-type symmetry for which k > 0 indicate both CH3OH A− and
CH OH A+3 levels. The A- type levels are further split into doublets indicated by + and
symbols. The arrows in both A-type and E-type diagrams show observed methanol maser
transitions. The arrow in the A-type diagram shows the 6.7 GHz maser transition (De Buizer,
2000).
The methanol molecule’s rotational, vibrational and electronic eigenstates are characterised by
quantum numbers with selection rules that only allow transitions from A-type to A-type and
E-type to E-type. Maser action only occurs on these allowed transitions as guided by quantum
mechanical selection rules for transitions between two energy states. A-type to E-type transi-
tions and vice versa are forbidden. Rotation states for both A-type and E-type are characterised
by Jk where J is total angular momentum and k is the projection of the angular momentum
on the molecular symmetry axis. A-type energy levels are characterized by quantum numbers
25
 University of Ghana http://ugspace.ug.edu.gh
k > 0 with an example being the 6668 MHz emission line which occurs between the 51 and 60
levels. The E-type energy levels are characterized by quantum numbers ±k with an example
being the 12178 MHz emission line for k being negative and the transition occurring between
the 20 and 31 levels. In the 51− 6 +0 A transition of the 6668 MHz and the 20− 31 E transi-
tion of the 12178 MHz methanol lines shown in Figure 1.12 radiative pumping occurs. The
methanol molecule in its ground state (J, k = 0) on absorption of an infra-red photon is excited
to the (J = 1, k = 0) or (J = 1, k = 1) states. Then the molecule decays to the metastable
(J = 1, k = 0), (J = 2, k = 2) and (J = 2, k = 0) levels before cascading back down. If the
infra-red pumping rate is faster than the spontaneous decays population levels in low energy
states are efficiently transferred to more excited states.
1.3.1.4 Saturation
For a maser medium which has been pumped to inversion, some of the atoms excited to the
upper level fall back to the lower level through stimulated emission and spontaneous decay. The
rate equation for the population of the upper state of the two level system contains a contribution
from absorption given by
∣
dn2 ∣∣∣ = B12n1K (1.35)dt abs
a contribution from stimulated emission given by
∣
dn2 ∣∣∣ =−B21n2K (1.36)dt stim
and a contribution from spontaneous emission given by
∣
dn2 ∣∣∣ =−A21n2 (1.37)dt spon
with similar considerations for the lower level. The populations of the levels n1 and n2 in this
system are thus governed by the rate equations
dn1
= n2A21 +(n2−n1)B21K +P1(n−n1−n2)−Γdt 1
n1 (1.38)
26
 University of Ghana http://ugspace.ug.edu.gh
and
dn2
=−n2A21− (n2−n1)B21K +P2(n−n1−n2)−Γ2n2 (1.39)dt
where n is the total number density of the molecular species involved in the maser, P1 and P2
are the pump rates per molecule into level 1 and 2, Γ1 and Γ2 are the loss or damp rate for level
1 and 2 and the saturation intensity
1 ∫
K = IνdΩ (1.40)4π
is the number of photons crossing a unit volume per unit time from all directions. For a photon
emitted spontaneously, the inverted medium serves as the gain medium as it travels through it
(Cohen, 1989). The negative optical depth of a maser amplifies it exponentially according to
Iν ∝ eτν (1.41)
Initially the exponential growth dominates as the pumping mechanism maintains the inverted
population by stimulated emission against the increasing losses. The fluence increases expo-
nentially with distance as well. The maser is said to be unsaturated and is characterised by a
narrow linewidth, rapid variability and a compact size. As population inversion is dependent on
the differences between the pump and the loss rates and the losses by the stimulated emission,
a change in the population levels that brings n2 close to n1 means absorption and emission go
to zero. Thus the maser itself affects the population inversion and is then said to be saturated
when
Γ
K 2= (1.42)
2B21
Figure 1.13 is an illustration of the behaviour of the maser when it is unsaturated for the expo-
nential growth, and saturated where the maser growth becomes linear.
27
 University of Ghana http://ugspace.ug.edu.gh
Figure 1.13: Maser saturation plot adopted from Cohen (1989) in which for an unsaturated
system the maser growth is exponential until such a distance demarcated with the broken line
when it becomes saturated and the growth becomes linear.
1.3.1.5 Polarization
In the presence of magnetic fields some hyperfine levels of paramagnetic molecules further
split into sublevels which give rise to additional line components. The extent of this splitting
(Zeeman effect) gives an indication of the magnetic field strength and the direction of the field
towards or away from the observer. The lines corresponding to the splitting exhibit polarization
effects, resulting in some maser polarization being as high as 100% (Cohen, 1989).
Figure 1.14 shows Zeeman splitting of the hydroxyl 2π3/2 hyperfine levels that gives rise to
polarization of the 1665 MHz transition (Garcia-Barreto et al., 1988). The π transitions occur
when 4ml = 0 giving rise to linear polarisation. Transitions in which 4ml =±1 result in the
σ components and give rise to circular polarisation. Here the electric field vector at a fixed
location is constant in magnitude but describes a circle in the plane perpendicular to the wave
vector (Jackson, 1999). The +σ component gives the left circular polarisation and the −σ the
right circular polarisation.
28
 University of Ghana http://ugspace.ug.edu.gh
Figure 1.14: The Zeeman splitting of the hydroxyl 2π3/2 hyperfine levels that gives rise to the
1665 MHz transition polarizations. Adapted from Garcia-Barreto et al. (1988).
1.4 Variability of masers
It has been shown (Sullivan and Kerstholts, 1976; Caswell, Vaile and Ellingsen, 1995; Macleod
and Gaylard, 1996; Goedhart et al., 2003, 2004; Green et al., 2012) that some masers in mas-
sive star forming regions show variations in their flux density. These flux variations can be
a flare or a fluctuation in amplitude over a period ranging from days to several years (Sul-
livan and Kerstholts, 1976; Macleod and Gaylard, 1996). The behaviour of the fluctuations
may be characterised as unchanging, monotonic increase or decrease in flux density, sporadic,
aperiodic, quasi-periodic and periodic with the change in amplitude either decreasing or in-
creasing. Some sources show a slow change in flux density (Szymczak et al., 2001) over the
period of observation whereas others show rapid change over short durations (Goedhart et al.,
2004). Some sources are even reported to exhibit coherent variations in all their maser features
whereas others only vary in a particular maser feature (Sullivan and Kerstholts, 1976).
In establishing how common the 6.7 GHz methanol variabilities are in massive star forming
regions, Goedhart et al. (2004) monitored a sample of 54 sources and reported 55% of the
372 maser features to show significant variations, confirming the results of Caswell, Vaile and
Ellingsen (1995) but with a higher percentage. Caswell, Vaile and Ellingsen (1995) studied
29
 University of Ghana http://ugspace.ug.edu.gh
variations in the intensity of methanol masers at 6.7 GHz and 12.2 GHz and found independent
correlated variation in some features.
Closely associated with methanol masers in massive star forming regions are the OH main-
line masers (1665 MHz and 1667 MHz). The spatial relationships between OH masers and
class II methanol masers have been mapped (Caswell, Vaile and Forster, 1995; Walsh et al.,
1998; Minier et al., 2001). These close associations and expectations from numerical mod-
elling (Cragg et al., 2002) suggest that both maser species are pumped by infra-red radiation
from the surrounding dust warmed by the central proto-star, so a level of similarity in behaviour
could be expected. Various changes in the degree of circular polarisation of several masers have
been reported (Szymczak et al., 2001). Investigations of the variability of the left circular po-
larisation (LCP) and right circular polarisation (RCP) of the 1665 MHz OH masers in some
complexes found 8 of the 10 exhibiting variabilities on the time scale of weeks, months and
even years (Clegg and Cordes, 1991). Monitoring of the 1.6 GHz hydroxyl and the 6.7 GHz
methanol masers in the massive star forming region G351.78− 0.54 from 1993 to mid 1995
showed monthly variation superimposed on long term changes varying for years in the LCP
(Macleod and Gaylard, 1996) and studies of 13 sources over the period from 1970 to 1971
found the mainline OH masers to vary in all four Stokes parameters on short time scales com-
pared to the satellite lines (Sullivan and Kerstholts, 1976). In contrast, a monitoring campaign
(Green et al., 2012) on the 1.6 GHz masers in G12.889+0.489 showed no significant variation
in the polarisation properties.
1.5 G9.62+0.20E
The massive star forming complex G9.62+0.20 has been found to be located at 5.2 kpc from
the Sun (Sanna et al., 2009, 2015). Within this complex shown in Figure 1.1 are a number
of HII regions at various stages of development (Hofner et al., 1996, 2001; Testi et al., 2000;
De Buizer, 2003; Liu et al., 2011) named A to I in increasing order of compactness (Garay
et al., 1993; Testi et al., 2000). The component E has been been detected at cm wavelengths
from 3 cm to 7 mm (Linz et al., 2005). This has been interpreted (Hofner et al., 1996) to be a
30
 University of Ghana http://ugspace.ug.edu.gh
young star surrounded by a very small UCHII region which on the Kurtz-Franco classification
(Kurtz and Franco, 2002) is dense and compact. Component E also harbours sources that emit
some hydroxyl maser lines (Caswell, 1998; Fish et al., 2005; Hofner et al., 1996), water maser
lines (Hofner et al., 1996; Testi et al., 1998; Forster and Caswell, 2000; Farmer, 2014) and
methanol maser lines (Menten, 1991; Goedhart et al., 2003) of which the 6.7 GHz transition is
the strongest in the Galaxy (Caswell, 1998; Sanna et al., 2015).
1.6 Periodicity of the 6.7 GHz methanol masers in G9.62+0.20E
The 6.7 GHz methanol transition in G9.62+0.20E was reported to show periodic changes in
its flux density in the form of regular flares with a period of 244 days (Goedhart et al., 2003,
2004). Figure 1.15 is a plot showing the periodic flares of the 6.7 GHz masers in G9.62+0.20E
(Goedhart et al., 2004).
Figure 1.15: Time series of the peak velocity channel of the 6.7 GHz masers in G9.62+0.20E
(Goedhart et al., 2004).
G9.62+ 0.20E is of interest because of the outstanding questions posed by the source of
the variability of the flux density of its associated 6.7 GHz and 12.2 GHz masers. Ten years of
monitoring of the regularly varying source using the Hartebeesthoek 26 m telescope produced
more results to confirm the periodic variation (Goedhart et al., 2014). The source of the flares
or variabilities of the maser’s intensity has been interpreted in a number of ways with several
models depending on the location the maser is speculated to be sited. Three of the models
31
 University of Ghana http://ugspace.ug.edu.gh
invoke the presence of a binary system whereas two invoke the presence of an independent
pulsating star.
1.6.1 Pulsating star models
The pulsating star model is derived from the swollen star model, in which a growing proto-
star under rapid mass accretion rate Ṁ∗ & 10−3 M −1 yr may become unstable and pulsate
before reaching the ZAMS (Hosokawa and Omukai, 2009; Hosokawa et al., 2010). In this
model it is the IR pumping radiation that is modulated and results in the periodicity of the
masers. Inayoshi et al. (2013) invoke the pulsational instability of bloated proto-stars in general
with high accretion rates to explain the observed maser periodicity and predict that the period
depends on the adopted accretion rates and gets longer with increasing accretion rates. Sanna
et al. (2015) also invoke the presence of an independent pulsating massive young proto-star
with high accretion rate of 4× 10−3 M −1 yr (Liu et al., 2011) and the superposition of two
independent IR sources (radiation field ) which provides the pumping mechanism for the maser
to explain variable maser behaviour in G9.62+0.20E.
1.6.2 Accretion disk models
The binary system models invoke the presence of a circumbinary disk to explain the dust tem-
perature variation which is assumed to cause the periodic variability. In one such model, the
observed variabilities in formaldehyde and methanol masers are explained as due to periodic
accretion of material from a circumbinary disk in a young eccentric binary system that heats
the dust and changes the pumping radiation field (Araya et al., 2010). Here the material from
the circumbinary disk is accreted onto the individual proto-star, heating the dust and increasing
the infra-red radiation field resulting in the large maser gain which leads to the high amplifi-
cation. This model follows predictions of the smoothed particle hydrodynamics simulations
(Artymowicz and Stephen , 1996) to arrive at this result.
Parfenov and Sobolev (2014), however, propose periodic emission from rotating spiral
shocks in the gaps of a circumbinary accretion disk around young binary stars to heat up the
dust. The variation in dust temperature is caused by radiation from the bow shock illuminating
32
 University of Ghana http://ugspace.ug.edu.gh
the dust which may again change the infra-red pumping field.
1.6.3 Colliding wind binary model
Methanol masers associated with UCHII regions are excited by strong external radiation from
hot dust (pumped by infra-red radiation) associated with the free-free emission from young HII
regions. The colliding wind binary (CWB) model has been employed to explain the observed
periodic variabilities of such masers (van der Walt et al., 2009; van der Walt, 2011b,a; van den
Heever et al., 2019). Unlike the previous models, here the seed photons are provided by the
HII region with high emission measure (Slysh et al., 2002; Sobolev et al., 2007).
In this model it is the seed photons that are periodically modulated causing the observed
periodicity. The model involves two massive OB stars in an eccentric binary system as in Fig-
ure 1.16 of which one or both produces large amounts of ionising radiation surrounding both
to power an HII region. The interaction of their stellar winds produces shocked gas which
emits more ionizing photons. A pulse of ionising photons is produced as the binary system
approaches periapsis. The pulse then propagates to the ionisation front of the HII region unat-
tenuated. This causes an increase in the electron density at the ionisation front resulting in the
subsequent increase in the free-free emission from that part of the HII region. The HII region
undergoes recombination and returns to its initial stage after the transit of the pulse. The maser
assumed to be projected against the edge of the HII region is expected to respond to the changes
in the free-free seed photons.
Figure 1.16: A toy model of the colliding wind binary model (van der Walt, 2011a).
33
 University of Ghana http://ugspace.ug.edu.gh
The colliding wind binary model therefore suggests that such periodic sources could be
important pointers for finding massive binary stars since massive stars have a high frequency
of forming in binaries and multiples (Zinnecker and Yorke, 2007). Massive close binary ob-
jects going supernova may form binary black holes or neutron stars. When the binary ob-
jects merge they generate gravitational waves as recently detected by the Laser Interferometer
Gravitational- Wave Observatory (LIGO) (Abbott et al., 2017, 2009a,b). Observational studies
of the early stages of high mass stars provide insight into the problem of how they form.
1.7 Thesis goals and overview
Closely associated with methanol masers in newly formed massive star forming regions are
hydroxyl masers, both of which are proposed to be excited by a common infra-red radiation
mechanism (Cragg et al., 2002). The CWB model suggests that these radiatively pumped
masers should vary together. However, the exact location of these masers in relation to the
proto-star is unknown presently. Sanna et al. (2015) used high resolution separate VLA and
VLBA astrometry to show the positions of masers relative to the 43 GHz UCHII region in
G9.62+0.20E. The results of the VLA continuum (UCHII) at 43 GHz does not map the actual
seed photons in the continuum at 1.6 GHz and 6.7 GHz.
This Ph.D. project seeks to find the correlation between the polarisations of the 1.6 GHz OH
and 6.7 GHz methanol masers, and to improve the picture of the positional relationship between
the maser emission and the HII regions of the high mass star by investigating the natal environ-
ment of G9.62+ 0.20E using hydroxyl and methanol maser emissions that uniquely trace the
regions within which massive stars are born and reside. In particular, the project utilizes the
26 m HartRAO dish to monitor the highly polarised mainline 1.6 GHz hydroxyl transitions and
the associated 6.7 GHz methanol maser emission in G9.62+0.20E to:
(a) investigate the variability of the left circular polarisation (LCP) and right circular
polarisation (RCP) components of the 1665 MHz and 1667 MHz OH maser lines.
(b) compare the results in (a) with the 6.7 GHz methanol maser line variabilities to inves-
tigate the relationship between the flaring behaviour of the methanol maser and the OH
34
 University of Ghana http://ugspace.ug.edu.gh
masers.
To understand the mechanism causing the maser flaring or variability in (a) and (b) above
the capability of e-MERLIN’s 150 mas resolution at L-band and 40 mas resolution at C-band
is employed to examine simultaneous high resolution images of the 1.6 GHz hydroxyl and
6.7 GHz methanol maser spots and their radio continuum emissions arising from G9.62+0.20E
to:
(c) measure the L-band and C-band continuum to high accuracy.
(d) determine the relative positions of the variable and non-variable maser spots in (a)
and (b).
This thesis thus investigates the relationship between the OH and CH3OH maser emissions and
the relationship between the masers and the HII region associated with them. The rest of the
thesis is organised as follows.
Chapter 2 introduces the techniques for reducing the single dish data of the 1.6 GHz main-
line OH polarisations and the 6.7 MHz methanol line in G9.62+ 0.20E. The differences and
similarities in the variabilities are discussed and an attempt is made to explain these variable
behaviours within the colliding wind binary model. Chapter 3 gives a general overview of
interferometry and the data reduction procedure for the high resolution interferometric data
of G9.62+ 0.20E at L-band. The study is continued in Chapter 4 by finding the positions of
the 6.7 GHz methanol maser spots in G9.62+ 0.20E relative to the 6.5 GHz radio continuum
emission using the high resolution observations. The concluding chapter gives a summary of all
inferences made from Chapter 2, Chapter 3 and Chapter 4 and how they all fit into the general
massive star formation theory.
35
 University of Ghana http://ugspace.ug.edu.gh
 University of Ghana http://ugspace.ug.edu.gh
CHAPTER 2
MONITORING THE POLARISATIONS OF THE HYDROXYL
MASERS IN THE PERIODIC METHANOL MASER SOURCE
G9.62+0.20E
Modelling the conditions necessary for maser formation has revealed that OH and methanol
masers are likely pumped by the same excitation mechanism. Hence, a spatial association
between the two species is expected (Cragg et al., 2002). Monitoring of the total intensity
of the OH mainline masers in G9.62+ 0.20E with KAT-7 found variations similar to those
observed in the 6.7 GHz and 12.2 GHz methanol transitions (Goedhart et al., 2019). However,
a drop in flux density of the OH masers which coincided with the rise of the 12.2 GHz methanol
transition was found (Goedhart et al., 2019). Polarisation variations in G9.62+0.20E were not
reported. However, the left and right circular polarisations of the OH mainline masers are
thought to originate from different regions in a molecular cloud and differences in behaviour of
the polarization components of the OH mainline masers are therefore expected.
Presented in this chapter are the results of the observation, reduction and analysis of data
obtained from the monitoring of the polarisation of the 1665 MHz and 1667 MHz hydroxyl
masers and the regularly varying 6.7 GHz methanol maser in G9.62+ 0.20E. The aim is to
find correlated variation between the 6.7 GHz methanol masers and the left and right circular
polarisations of the OH masers. Such information is valuable for identifying the source of the
periodic variation observed in some maser species. The monitoring campaign was instituted
with the 26 m HartRAO radio telescope in 2011.
Section 2.1 gives brief explanations of the observational data collection techniques em-
ployed. In section 2.2 detailed descriptions of radio astronomical spectroscopic data reduction
37
 University of Ghana http://ugspace.ug.edu.gh
are given, and section 2.3 presents the results of the reduction. The results are discussed in
section 2.4.
2.1 Observational techniques
Data for both OH and methanol maser observations were obtained using the 26 m Harte-
beesthoek Radio Astronomy Observatory (HartRAO) microwave telescope which is an equato-
rial mount with an 85 foot Cassegrain dish design, built by Blaw Knot in 1961. Table 2.1 gives
the specifications for the telescope.
Table 2.1: Specification for the 26 m HartRAO telescope.
Parameter Specification
Diameter (m) 25.9
Focal ratio ( f/D) 0.424
Surface tolerance (mm rms) 0.5
Wavelength limit (cm) 1.3
Feed system Cassegrain
Pointing resolution (deg) 0.001
Repeatability (deg) 0.004
Slew rate on each axis (deg/s) 0.5
Depending on the purpose of the observation different techniques were employed. The amount
of power measured by a telescope consists of noise from the atmosphere and radiation from
the radio source. To measure the power due to the radio source the noise is subtracted from
the total measured power. This is achieved through switched observations of the source. The
switching technique takes a number of forms: position, beam, load and frequency switching.
For spectral line and some continuum observations, position, frequency and load switching
are used whereas beam switching and drift scans are used for continuum observations of point
sources.
Two monitoring windows, a regular and an intensive window, were employed for the col-
lection of the data. The regular monitoring data had been taken by HartRAO staff and archived
whereas the intensive window data were the product of a follow up proposal for monitoring of
the source led by Benedicta Woode which was accepted by HartRAO. The regular monitoring
window spanned the period 2011 day 294 to 2015 day 189 for the observation at 1665 MHz
38
 University of Ghana http://ugspace.ug.edu.gh
and 2014 day 066 to 2015 day 189 for the observation at 1667 MHz. For the 1665 MHz line,
144 observations were recorded whereas for the 1667 MHz line 66 observations were recorded.
The 6668 MHz methanol maser data for the regular monitoring window were supplied by Dr.
Sharmila Goedhart from her monitoring campaign data taken at HartRAO. The intensive mon-
itoring data were taken during the 2017 January to April flare window which covered day 007
to day 130 for the observation of both OH and methanol emissions. A total of 48 observations
were taken for the methanol maser line, 22 observations were taken for the 1665 MHz OH line
and 30 observations were taken for the 1665 MHz OH line.
2.1.1 Frequency switching
Observations were carried out with the spectrometer configured to produce 1024 channels
across a 0.25 MHz bandwidth in both LCP and RCP. The instrumental gain response as a
function of frequency, which is the bandpass, changes slowly with time as the observation pro-
gresses, and correlator artefacts as well as standing waves add to this variation. Provision is
made for this in HartRAO by observing either in position switching mode or frequency switch-
ing mode, both of which involve the subtraction of a reference spectrum (off-source) from the
signal (on-source) observation in order to remove all common effects.
For position switching the reference spectrum is taken by pointing the telescope a few beam-
widths away from the source. The disadvantage of this method is that half the observation time
is spent off the target, and in crowded regions it may not be possible to get away from other
emission sources. In frequency switching, the spectral line occupies a narrow band of the
frequency range of the instrument. To generate the signal and the reference spectral pair, the
central observing frequency is shifted by the local oscillator slightly on and off the spectral
line such that the line emission falls on different sides of the band. This way the common
contributions from both the receiver and the atmosphere are recorded in both frequency settings,
which is a more efficient use of telescope time compared to position switching. This technique,
however, is not always possible if the frequency range of the line emission is more than half
the spectrometer bandwidth.
The frequency switching technique was employed in monitoring the source at intervals of
39
 University of Ghana http://ugspace.ug.edu.gh
1–2 weeks during the regular monitoring window and the monitoring interval was decreased
to 2–3 days during its projected flaring phase. Daily monitoring was done during the intensive
monitoring window and as the source entered its quiescent phase the observations were reduced
to 2–3 day intervals.
2.1.2 Drift scans
The contraction and expansion of a radio telescope which comes about as a result of changes in
atmospheric temperature conditions as well as the earth’s gravitational force as it points away
from zenith causes the deformation of the telescope’s structure. This deformation introduces
errors in the measurements of flux densities. At the shorter wavelengths, that is, ≤ 6 cm for
the HartRAO 26 m telescope, the main beam becomes small resulting in slight variations in
its pointing, thus the full width at half maximum of the main beam decreases with increasing
frequency. To accurately measure the peak flux of the source, drift scans at the expected north
and south half-power points as well as the centre were taken. For the methanol observations
short, one minute pointing checks to the half power cardinal points (north, south, east, and
west) followed by two on-source centre position observations were carried out. These were
followed by a pair of full length five minute scans on the central position. Pointing checks were
not required for the OH observations as the beam is larger. For calibration purposes drift scans
through the chosen calibrators were used. This is done by parking the telescope at a coordinate
ahead of the calibrator’s current location along its path. The moving source then drifts over the
telescope beam.
Table 2.2 summarizes the observing parameters for 1665 MHz and 1667 MHz frequencies and
Table 2.3 gives the observing parameters for the 6668 MHz frequency.
2.2 Data reduction
Reduction of the data was performed with a Python script optimized for the processing of the
spectral line data. A plot of antenna temperature against radial velocity generated the raw
spectra which show the details of the bandpass. For example, Figure 2.1 shows the spectrum
40
 University of Ghana http://ugspace.ug.edu.gh
Table 2.2: Observing parameters for G9.62+0.20E at 1665 MHz and 1667 MHz.
Parameter Specification
Rest frequency at 1665 MHz (MHz) 1665.401840
Rest frequency at 1667 MHz (MHz) 1667.359030
Half power beamwidth (arcmin) 28.5
System temperature (K) 39
Number of spectral channels 1024
Polarizations LCP + RCP
Correlator bandwidth (MHz) 0.25
Frequency switch offset (MHz) 0.125
Channel width (kHz) 0.244
Vlsr (km/s) 2.5
Table 2.3: Observing parameters for G9.62+0.20E at 6668 MHz.
Parameter Specification
Rest frequency (MHz) 6668.518000
Half power beamwidth (arcmin) 7.1
System temperature (K) 57
Number of spectral channels 1024
Polarizations LCP + RCP
Correlator bandwidth (MHz) 1.0
Frequency switch offset (MHz) 0.5
Channel width (kHz) 0.976
Vlsr (km/s) 2.5
for an observation on day 066 of 2014 during the regular monitoring window. This observation
is used to illustrate the single dish reduction procedure adopted for both monitoring windows.
The total power output in LCP shown in Figure 2.1 comprises the maser line visible at
2.5 km/s, the sky level at about 45 K, the bandpass profile which falls off drastically near the
edges, and the correlator artefact in the form of a spike at about 13 km/s. The spectrum’s
velocity axis has been calibrated to radial velocity (Vlsr) scale and Doppler tracking was also
applied at HartRAO.
2.2.1 Bandpass correction
The bandpass effect of the spectrum, which is the most dominant component from the instru-
ment, was corrected by subtracting the sky emission (∼ 45 K) and the edge of the bandpass
using the frequency switched spectra. The frequency switched spectra consist of a reference
spectrum and signal observations. Figure 2.2 shows the spectrum of the output of the correlator
41
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.1: The raw total power spectrum from the 1024 channel spectrometer showing the
bandpass and the maser signal in the left circular polarisation (LCP).
which constitutes with Figure 2.1 a frequency switched pair. The left half of the spectrum in
Figure 2.1 and the right half of the spectrum in Figure 2.2 share common velocity channels.
Similarly, the right half of the spectrum in Figure 2.1 and the left half of the spectrum in Figure
2.2 also share common velocity channels but Figure 2.2 has negative values. A channel by
channel subtraction of the spectrum in Figure 2.2 from the spectrum in Figure 2.1, and vice
versa, returns the two spectra shown in Figure 2.3.
42
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.2: The second spectrum of the frequency switched pair in LCP showing the bandpass
with the spectrometer artefact at about −7 km/s and maser line emission offset by half the
bandwidth to the right of the spectrum.
43
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.3: The first observation of the frequency switched pair and the second observation of
the frequency switched pair both showing positive and negative velocity maser features offset
by 22.5 km/s.
The first resulting spectrum has a maser feature at 2.5 km/s and a feature at a negative
velocity offset by 22.5 km/s (0.125 MHz) from the positive velocity feature as shown in the
first observation in Figure 2.2. Similarly, the second resulting spectrum has a positive velocity
maser feature at 2.5 km/s and a negative velocity feature offset by 22.5 km/s ( 0.125 MHz) as
in the second observation in Figure 2.3. The two spectra were combined as shown in Figure
2.4 and the average spectrum over the common velocity range −8.75 km/s to 13.75 km/s is
shown in Figure 2.5. It can be seen that the continuum level is now close to zero. The procedure
also removes the central spike from the spectrometer but leaves residual frequency dependent
effects, which manifest as a slope in the baseline. In some cases ripples appeared in the base-
lines of the spectra but were also removed with this procedure. These ripples were caused by
the blocked aperture of the HartRAO telescope which produces standing waves between the
secondary and the main surface.
44
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.4: Combined observation of the frequency switched pair.
45
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.5: Averaged frequency switched spectra showing the baseline level removed.
2.2.2 Radio-frequency interference excision
The spectra from some of the L-band datasets were corrupted with strong radio-frequency
interference (RFI), which manifested as spikes in the spectrum. Figure 2.6, a spectrum from
the observation on day 092 of 2014 for the regular monitoring window shows an example of
such data with spikes at −7 km/s and 14 km/s. The negative RFI at 2.5 km/s and −5 km/s
result from the bandpass correction process using the frequency switched pairs. In most of
the infested spectra from the dataset the radio-frequency interference spikes were randomly
46
 University of Ghana http://ugspace.ug.edu.gh
distributed. For these corrupted data a spike removal exercise was employed to remove the
spikes from the spectra.
Figure 2.6: Spectrum showing persistent radio-frequency interference manifesting as spikes
and dips.
The RFI spikes were removed by first masking the maser lines. A median absolute deviation
(MAD) was then calculated on the rest of the emission-free spectrum. The MAD was multiplied
by 1.48 to return an estimate of the standard deviation σ assuming normally distributed data.
Any deviation greater than the estimated threshold by 3σ was treated as an outlier and replaced
with a local median calculated over 5 channels. Figure 2.7 shows the spectrum in Figure 2.6
with RFI adequately excised.
47
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.7: The spectrum in Figure 2.6 with RFI spikes and dips removed.
2.2.3 Smoothing of spectrum
In some instances the RFI was so strong that it resulted in ringing of the baseline. The spectrum
was smoothed with a Hanning window with a width of 6 channels. This improves the SNR by a
√
factor of 6. The Hanning window was used because it uses a smoothening kernel which does
not reduce the number of channels. Figure 2.8 shows the result of performing the smoothing
procedure on the spectrum shown in Figure 2.7. The ringing in the unsmoothed spectrum is
substantially decreased by the procedure.
48
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.8: Ringing in the baseline smoothed out with a Hanning window algorithm.
2.2.4 Baseline flattening
The residual baseline variation was removed by fitting a polynomial to the baseline of the
bandpass. First the two maser line ranges which occur at 0 km/s–3 km/s and 5 km/s–7 km/s
for 1665 MHz and 0 km/s–3 km/s and 6 km/s–8 km/s for 1667 MHz were masked. A third
order polynomial was fitted to the baseline as shown in Figure 2.10. This provided a better
improvement of the systematics against the other baseline fits and was also of an order low
enough as not to distort the emission features in the spectrum. A second order fit did not
49
 University of Ghana http://ugspace.ug.edu.gh
capture the trend in the data, nor correct adequately for the variation in the sky level. A higher
order of four over-fitted the data, fitting most points in the data and introducing ripples under
the source as evidenced in Figure 2.9.
Figure 2.9: A fourth-order polynomial fit of a green curve to the spectral baseline showing an
uneven fit to the bandpass. The blue coloured part of the spectrum shows the masked maser
line and the orange part of the spectrum shows the baseline.
Figure 2.10: The top plot illustrates a third-order polynomial fit of a green curve to the spectral
baseline with the masked range being the start and end velocity of line emissions shown in
blue.The bottom plot shows the results of the subtraction of the baseline fit from which flattens
the bandpass.
A plot of the resulting spectrum, presented in Figure 2.11, shows removal of the slope,
50
 University of Ghana http://ugspace.ug.edu.gh
sky (level) and other contributions with the best fit for the data and baseline, a third order
polynomial.
Figure 2.11: The spectrum showing the flattened baseline of both LCP and RCP with the mean
signal at 0 K antenna temperature.
2.2.5 Pointing correction
To calculate the pointing correction factor, short five point scans on the source (in this case
the methanol maser) were taken by offsetting half of the beam width to the cardinal points
(north, south, east and west) from the requested source center. If the pointing is correct half the
power at the center is recorded at the half power points but if the pointing is off, more power
51
 University of Ghana http://ugspace.ug.edu.gh
is recorded at some cardinal points than others. Thus for a perfect pointing model, the peak of
the scan is recorded at zero right ascension (RA) offset. The cross-section P of the main beam
of the telescope can be approximated by a Gaussian profile given by
( ( ) )
δθ 2
P = P0 exp − ln2 (2.1)b
Here P represents the main lobe of the antenna beam pattern, δθ is the offset between the
source and the centre of the beam and b is the half power beam width. Hence a Gaussian fitted
to the main beam data (peak of each maser line) as in Figure 2.12 measures the amplitude
(maximum power) of the peak for each cardinal point.
Figure 2.12: Peak power of the spectra taken at the half power point of cardinal points (north,
south, east and west) and on the source center.
If the telescope has a good pointing model the peak of the Gaussian is expected close to the
centre of the scan. Given the power measured at the half-power point north, south, east, west
and on the center of the beam, the pointing offset for the north-south direction as well as that for
the east-west direction which is the measured signal amplitude as a function of offset position
(peak) was calculated. An example of a pointing offset calculation in degrees for north-south
and east-west directions for an observation is shown in Figure 2.13. The pointing offset values
are the pointing correction factors applied to the on-source amplitude. The average pointing
correction factor for the 6668 MHz data was 1.067.
52
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.13: The left plot shows the north-south pointing offset with an offset of 0.015deg from
the center and the right plot displays the east-west pointing offset of 0.017deg from the center.
2.2.6 Flux density calibration
To calibrate the flux density of the maser lines in the spectra, three drift scans on the flux
calibrators were taken, one through the expected center of the beam and the others through the
expected north and south half-power points respectively. The amplitudes were calculated by
approximating each scan to a Gaussian and fitting a Gaussian plus a linear baseline to the main
peak of each drift scan as shown in Figure 2.14. The Gaussians used for the fitting are:
( ( ) )
δθ 2
Tcenter = T exp − ln2 (2.2)b
( ( ) )
δθ +b 2
Tnorth = T exp − ln2 (2.3)b
( ( ) )
δθ −b 2
Tsouth = T exp − ln2 (2.4)b
where Tcenter is the peak amplitude at the center, Tnorth is the peak amplitude at the north and
Tsouth is the peak amplitude at the south half-power points on the beam. T evaluates to the
antenna temperature which is used to obtain a conversion factor between the signal of the
source in units of antenna temperature TA and flux density S (in Jy ). The conversion factor, the
53
 University of Ghana http://ugspace.ug.edu.gh
point source sensitivity (PSS), which is the flux density in janskys that produces one kelvin of
antenna temperature change and measured in Jy/K, is used to convert the spectrum intensity
measured in antenna temperature to flux density. The inverse of this is the gain G (Marr et al.,
2016). For an unresolved point source the relationship between antenna temperature TA and
flux density S is given by
S = 2kBTAAeff (2.5)
where kB is Boltzmann’s constant and Aeff is the effective geometric area of the telescope. Thus
the PSS for each polarization of a polarized radio point source is given by
PSS = S/2KsTA (2.6)
where Ks is the correction factor for the source (Ks = 1 for a point source). The flux density Ssrc
of some source now becomes
Ssrc = PSScalTAsrc (2.7)
where PSScal is the PSS of the calibrator and TAsrc is the antenna temperature produced by the
source.
Three sources, 3C123, Hydra A and Virgo A, were observed for calibration purposes. Virgo
A is the brightest calibrator of the three and was used to calculate the PSS at 1665 MHz,
1667 MHz and 6668 MHz. The flux densities of the calibrators were found from published
scales (Ott et al., 1994).
The antenna temperature in each polarization was converted to flux density by multiplying
it with the PSS of the corresponding polarization for both 1665 MHz and 1667 MHz maser
lines. For the flux density of the methanol 6668 MHz line, however, the antenna temperature of
the two polarizations were added to get the total antenna temperature which was then multiplied
by the PSS to obtain the flux density of the calibrated spectrum. Figure 2.15 shows the flux
calibrated spectrum for a hydroxyl maser.
The calculated point source sensitivities are recorded in Table 2.4.
54
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.14: Fitting of a Gaussian beam plus linear baseline to the drift scan on Virgo A to
calculate the amplitude of the peak.
Table 2.4: Point Source Sensitivity values calculated for LCP and RCP.
Monitoring window PSS LCP (Jy K−1) PSS RCP (Jy K−1)
Regular window 6.1 5.2
Intensive window 5.1 5.2
2.3 Results
The averaged spectra, plots of range of variation of the flux density and time series of the hy-
droxyl masers in LCP and RCP for both transitions as well as the methanol maser for both reg-
ular and intensive monitoring windows are presented in this section. The OH profiles recorded
during the survey were bright in the LCP and faint in the RCP. Each profile was examined
visually and bad data were removed from the data set. RFI which had not been removed sat-
isfactorily and not located close to the maser line did not affect the maser line; hence, such
spectra were included in the data. Ninety four spectra representing 27% of the data were re-
moved from the analysis because some RFI persisted after the RFI excision procedure. The
root mean square (rms) noise over the line-free emission region was measured by masking the
line emission range and calculating the noise of the line-free channels. The average flux density
and rms noise levels in the spectra for the hydroxyl and methanol lines are tabulated in Table
2.5 and Table 2.6 for the different monitoring windows.
Generally the spectra for LCP were found to contain more flux density than the RCP for
55
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.15: Calibrated spectrum after correction factor PSS has been applied to the on-source
amplitude of the antenna temperature of the hydroxyl spectrum to covert it to flux density.
Table 2.5: Average flux density and rms noise for the regular monitoring window.
Frequency (MHz) Average flux density (Jy) Average rms noise (Jy)
1665 LCP 10.34 0.50
1665 RCP 8.48 0.41
1667 LCP 23.66 0.54
1667 RCP 3.24 0.44
6668 I 5500.00 74.00
both monitoring windows as can be seen in Table 2.5 and Table 2.6. This, of course, is so
because the OH maser is almost 100% polarised in the LCP (Reid and Moran, 1981). A com-
56
 University of Ghana http://ugspace.ug.edu.gh
Table 2.6: Average flux density and rms noise for the intensive monitoring window.
Frequency (MHz) Average flux density (Jy) Average rms noise (Jy)
1665 LCP 23.28 1.07
. 1665 RCP 6.11 0.41
1667 LCP 25.23 0.78
1667 RCP 2.87 0.47
6668 I 3500 82
parison of the average flux densities in both windows shows that for the 1665 MHz line, the
flux density for LCP increased whereas the flux density in the RCP decreased. The same trend
is seen for the 1667 MHz line though the increase in the 1665 MHz LCP flux density was about
a factor of two greater than the increase in the 1667 MHz.
2.3.1 Averaged spectra for regular monitoring
The averaged spectra for each polarization of the OH mainlines and the total intensity for the
methanol lines are displayed in Figure 2.16 through to Figure 2.20 for the regular monitoring
window. The averaged spectra are the combination of the median values at each spectral chan-
nel taken from the values of the monitoring window. Each spectrum of the maser line is made
up of a number of peaks which represent emissions from maser spots at different velocities. A
velocity range of 25 km/s spread from −10 km/s to 15 km/s was used to display the detected
features in each spectrum. Table 2.7 displays the velocity channels with OH maser emissions
that were detected and their corresponding flux densities for the regular window.
57
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.16: Averaged spectrum for the 1665 MHz maser line in LCP.
Figure 2.17: Averaged spectrum for the 1665 MHz maser line in RCP.
58
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.18: Averaged spectrum for the 1667 MHz maser line in LCP.
The prominent maser components in each spectrum lie in the range 0 km/s to 3 km/s for
the hydroxyl maser and those of the methanol maser lie in the range 0.5 km/s to 3 km/s. The
velocities that have no emissions correspond to the noise level in the maser spectra.
59
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.19: Averaged spectrum for the 1667 MHz maser line in RCP.
Figure 2.20: Averaged total intensity spectrum for the 6668 MHz maser line.
60
 University of Ghana http://ugspace.ug.edu.gh
Table 2.7: Detected OH maser velocity channels and their flux densities for the regular moni-
toring window.
1665 MHz Flux density (Jy) 1667 MHz Flux density (Jy)
Vlsr (km/s) LCP RCP LCP RCP
−2.5 0.7 − − −
−1.5 − − − 0.3
−1.0 − − 1 −
0.5 − − − 1.5
1.0 0.7 7 − −
1.5 12 − 25 3.8
2.0 − 4.9 − −
4.0 0.6 0.3 − −
4.2 − − − 1.2
5.0 − 0.3 0.5 0.4
6.0 1 0.6 0.7 0.1
7.0 − − 4 1
2.3.2 Averaged spectra for intensive monitoring
Figure 2.21 through to Figure 2.25 show the averaged spectra for both transitions of OH and
the methanol line for the intensive monitoring windows. In Table 2.8 the maser features of the
OH transitions with their corresponding flux densities for the intensive monitoring window are
presented.
61
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.21: Averaged spectrum for the 1665 MHz maser line in LCP.
Figure 2.22: Averaged spectrum for the 1665 MHz maser line in RCP.
62
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.23: Averaged spectrum for the 1667 MHz maser line in LCP.
Figure 2.24: Averaged spectrum for the 1667 MHz maser line in RCP.
63
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.25: Averaged total intensity spectrum for the 6668 MHz maser line.
Table 2.8: Detected OH maser velocity channels and their flux densities for the intensive mon-
itoring window.
1665 MHz Flux density (Jy) 1667 MHz Flux density (Jy)
Vlsr (km/s) LCP RCP LCP RCP
−2.5 1 − − −
−1.0 0.5 − 1 −
−0.5 − − − 0.5
0 − 0.3 − −
0.5 − − − 1.03
1.5 27 6 28 2.8
4.0 0.5 − − 1
5.0 − 0.5 − −
6.0 2 0.4 0.5 −
7.0 − − 4 0.5
2.3.3 Spectra from Caswell et al. (2013)
Plots of the OH mainline polarization spectra adapted from Caswell et al. (2013) are presented
in Figure 2.26 and Figure 2.27. This ATCA observation of the G9.62+0.20 region consisting
64
 University of Ghana http://ugspace.ug.edu.gh
of G9.619+0.193, G9.620+0.194 and G9.621+0.196 is covered by the beam for the Har-
tRAO telescope. Also presented in Table 2.9 are the maser emissions which fall in the velocity
range−10 km/s to 30 km/s. The flux densities recorded in the velocity channels are presented
in Table 2.9 as well. In the absence of individual polarisation spectra from this region these
plots will be compared to the results.
Figure 2.26: The 1665 MHz spectrum of the three regions G9.619+0.193, G9.620+0.194
and G9.621+0.196 in G9.62+0.20 from Caswell et al. (2013) showing the left circular polar-
isation, right circular polarisation and linear polarisation features.
Figure 2.27: The 1667 MHz spectrum of the three regions G9.619+0.193, G9.620+0.194
and G9.621+0.196 in G9.62+0.20 from Caswell et al. (2013) showing the left circular polar-
isation, right circular polarisation and linear polarisation features.
65
 University of Ghana http://ugspace.ug.edu.gh
Table 2.9: Detected OH maser velocity channels and their flux densities derived from Caswell
et al. (2013).
1665 MHz Flux density (Jy) 1667 MHz Flux density (Jy)
Vlsr (km/s) LCP RCP LCP RCP
−4.0 0.5 − − −
−2.5 0.5 − − −
−1.5 − − 0.2 −
−1.0 1 − − −
0.5 − − − 1.9
1.0 − 6 − −
1.5 − − 9 2.1
2.0 9 8 − 2.6
4.0 2 2 − −
4.2 − − − 2.6
5.0 − 0.2 − 0.2
6.0 4 4 1 0.1
7.0 − − 5 1
2.3.4 Comparison of the averaged spectra of OH for regular and inten-
sive monitoring window with Caswell et al. (2013)
Here the results for both the regular and the intensive monitoring windows are compared with
those reported by Caswell et al. (2013). For the regular monitoring window Table 2.7 derived
from the regular monitoring plots shows five major emissions for both polarisations of the
1665 MHz line and the LCP of the 1667 MHz line whereas two more emissions are recorded
in RCP for the 1667 MHz line. A comparison of these results with the data in Table 2.9 shows
that all emissions reported in 2013 were again detected except the emission at channel 1 km/s.
An additional emission at −1 km/s for the 1665 MHz line in LCP was detected whereas the
1665 MHz RCP shows no emission at 4 km/s, 2 km/s and 1 km/s but rather at 1.5 km/s. The
1665 MHz RCP also shows signals at−3.5 km/s and−1 km/s which are not spurious. For the
1667 MHz spectrum detections of all the emissions except that at 5 km/s were made for the
LCP whereas for the RCP two additional emissions, at −0.5 km/s and 4 km/s, were made but
no emissions were recorded at −1.5 km/s, 4.2 km/s and 5 km/s. The RCP of the 1667 MHz
line also shows signals at −6 km/s, −4.5 km/s, −5 km/s, −1.5 km/s and some emissions
occur at velocities greater than 10 km/s.
66
 University of Ghana http://ugspace.ug.edu.gh
Turning to the intensive monitoring window a comparison of Table 2.8 for the intensive
monitoring window with Table 2.9 shows that all emissions for the LCP of the 1665 MHz line
except the features 1.5 km/s and 1 km/s were detected with additional emissions detected at
−4 km/s and −1 km/s. All maser emissions in RCP were coincident with those of Caswell
et al. (2013); however, an emission detected at −4 km/s is not real. The maser emissions in
the 1667 MHz transitions were also coincident with those of Caswell et al. (2013) except the
emission at 5 km/s and a new emission which was detected at−1.5 km/s in the LCP. Emissions
coincident with those of Caswell et al. (2013) were recorded for the RCP at all channels except
at −1.5 km/s. An additional emission at 2 km/s was also recorded. Emissions were also
recorded at channels −6 km/s, −0.5 km/s and 10.5 km/s of the RCP component but these are
not real.
2.3.5 Comparison of the averaged spectra of methanol for the regular
and intensive monitoring window with Goedhart et al. (2004)
The flux density for the 6.7 GHz methanol line from G9.62+0.20E was first measured to be
4870 Jy (Menten, 1991). Almost two decades later the same telescope measured the flux to be
5240 Jy (Green et al., 2010). A flux density of 4375 Jy has also been recorded with HartRAO
(Goedhart et al., 2004). Table 2.10 shows the results for the detected emissions during the
regular and intensive monitoring of the 6.7 GHz methanol maser. A comparison of Table 2.10
with the detected emissions from Goedhart et al. (2004) presented in Table 2.11 shows that all
emissions were detected. The flux density in the main velocity channel 1.23 km/s was 4504 Jy
for the regular window and 4320 Jy for the intensive window whereas Goedhart et al. (2004)
recorded 4375 Jy suggesting an increase in the flux density over time.
67
 University of Ghana http://ugspace.ug.edu.gh
Table 2.10: Detected 6668 MHz maser velocity channels and their flux densities for both regu-
lar and intensive monitoring windows.
Regular window Intensive window
Vlsr (km/s) Flux (Jy) Vlsr (km/s) Flux (Jy)
−1.89 21 −1.89 25
−1.27 38 −1.27 44
−0.84 113 −0.84 94
−0.22 677 −0.22 521
1.18 3698 1.18 4300
1.23 4320 1.23 4504
3.03 46 3.03 44
3.73 30 3.73 26
5.26 76 5.36 79
6.36 26 6.36 27
Table 2.11: Detected 6668 MHz maser velocity channels and their flux densities derived from
Goedhart et al. (2004).
Vlsr (km/s) Flux (Jy)
−1.88 20
−1.21 30
−0.76 41
−0.42 34
−0.14 423
1.26 4375
1.94 121
3.12 68
5.31 62
6.43 13
2.3.6 Range of variation for regular monitoring
Table 2.5 and Table 2.6 give the average flux density in each polarization for the 1665 MHz
and 1667 MHz maser lines. Within each velocity channel of each polarisation a minimum and
a maximum flux density is recorded which gives an indication of the range of variability of
the flux density. Figure 2.28 to Figure 2.32 display the range of flux density variation for the
regular monitoring window.
68
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.28: Range of variation of flux density in the spectrum of 1665 MHz LCP. The blue
line represents the upper envelope, the green line shows the averaged spectrum, and the orange
line shows the lower envelope.
69
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.29: Range of variation of flux density in the spectrum of 1665 MHz RCP. The blue
line represents the upper envelope, the green line shows the averaged spectrum, and the orange
line shows the lower envelope.
70
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.30: Range of variation of flux density in the spectrum of 1667 MHz LCP. The blue
line represents the upper envelope, the green line shows the averaged spectrum, and the orange
line depicts the lower envelope.
71
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.31: Range of variation of flux density in the spectrum of 1667 MHz RCP. The blue
line represents the upper envelope, the green line shows the averaged spectrum, and the orange
line depicts the lower envelope.
72
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.32: Range of variation of flux density in the spectrum of 6668 MHz. The blue line
represents the upper envelope, the green line shows the averaged spectrum, and the orange line
depicts the lower envelope.
2.3.7 Range of variation for intensive monitoring
Figure 2.33 to Figure 2.37 show the range of variation for the intensive monitoring window.
73
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.33: Range of variation of flux density in the spectrum of 1665 MHz LCP. The blue
line represents the upper envelope, the green line shows the averaged spectrum, and the orange
line depicts the lower envelope.
74
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.34: Range of variation of flux density in the spectrum of 1665 MHz RCP. The blue
line represents the upper envelope, the green line shows the averaged spectrum, and the orange
line depicts the lower envelope.
75
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.35: Range of variation of flux density in the spectrum of 1667 MHz LCP. The blue
line represents the upper envelope, the green line shows the averaged spectrum, and the orange
line depicts the lower envelope.
76
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.36: Range of variation of flux density in the spectrum of 1667 MHz RCP. The blue
line represents the upper envelope, the green line shows the averaged spectrum, and the orange
line depicts the lower envelope.
77
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.37: Range of variation of flux density in the spectrum of 6668 MHz. The blue line
represents the upper envelope, the green line shows the averaged spectrum, and the orange line
depicts the lower envelope.
2.3.8 Comparison of the range of variation for regular and intensive mon-
itoring windows
During the regular monitoring campaign, the flux density of the 1665 MHz LCP component var-
ied between 8 Jy and 18 Jy as shown in Figure 2.28 whereas the flux density of the 1665 MHz
LCP component in the intensive monitoring window varied from 24 Jy to 30 Jy as seen in Fig-
ure 2.33. This shows an increase in flux density variation with the change in flux density being
almost two fold. The flux densities of the RCP components in the 1665 MHz line in Figure
2.29 and Figure 2.34, however, varied between a maximum of 10 Jy and a minimum of 6 Jy.
For the 1667 MHz spectra similar variation behaviour was recorded as for the 1665 MHz as
can be seen in Figure 2.30, Figure 2.35, Figure 2.31 and Figure 2.36. The methanol line from
78
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.32 and Figure 2.37 however shows the same flux density variation with a maximum
of 8000 Jy but a minimum of 5500 Jy for the regular monitoring window and 3200 Jy for the
intensive monitoring window.
2.3.9 Time series of some selected channels: regular monitoring
Figure 2.38 to Figure 2.42 present time series for the OH and methanol lines for the regular
monitoring window. Five time series corresponding to five velocity channels are plotted for
each transition. The first three time series are selected from the most varying velocity channels
at peak of the main maser line. Since some OH lines presented double peaked spectra care was
taken so as to include only velocity channels from the maximum of the main peak. The remain-
ing two velocity channels were selected from line-free channels. This was done to demonstrate
the extent of variation between the emission channels and line-free channels. Here also MJD
represent the day of observation using the mean Julian calender. For simplicity the first two
digits "24" are dropped.
79
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.38: Time series for selected velocity channels of the 1665 MHz LCP component in the
regular monitoring window. The features in the velocity channels 1.36 km/s, 1.45 km/s and
1.53 km/s follow a monotonic increase in flux density from 6 Jy to 18 Jy whilst the features in
the velocity channels 9.62 km/s and 9.84 km/s show no variation. The black + signs plotted
on the 1.36 km/s velocity channel show the times the flares are expected.
80
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.39: Time series for selected velocity channels of the 1665 MHz RCP component in
the regular monitoring window. The features in the velocity channels 1.66 km/s, 1.71 km/s,
1.75 km/s, 2.72 km/s and 2.94 km/s only show sporadic variation in flux density from 4 Jy to
7 Jy. The black + signs plotted on the 1.66 km/s velocity channel show the times the flares are
expected.
The time series show a variety of behaviour. A feature considered for the test of variation
must have a flux density greater than 3 times the noise level. A feature is said to be varying
if the change in amplitude above the 3σ level is recorded. Variations were not recorded in all
features but were limited to a handful of features with the strongest variation usually occurring
at peaks of the spectra. Variabilities above the noise level were not detected for weaker features
with low signal-to-noise ratios. For both transitions of the OH species the behaviour of the
time series in some selected channels of the LCP component exhibited both short term and
long term variations for the regular monitoring period. In Figure 2.38 and Figure 2.39 which
show the behaviour of the time series in some selected channels for the 1665 MHz LCP and
81
 University of Ghana http://ugspace.ug.edu.gh
RCP components the LCP component exhibited significant variation in flux density whilst the
RCP component exhibited a form of sporadic variation in flux density. The 1665 MHz LCP
component showed three flares at 56325 MJD, 56900 MJD and 57100 MJD. The peak of the
flare at 57100 MJD in Figure 2.38 is not visible because of flagged data.
Figure 2.40: Time series for selected velocity channels of the 1667 MHz LCP component in
the regular monitoring window. Features in the velocity channels 1.58 km/s, 1.62 km/s and
1.67 km/s vary in flux density from 6 Jy to 24 Jy with two significant flares at 56900 MJD and
57100 MJD marked with + signs on channel 1.58 km/s. The features in the velocity channels
2.15 km/s and 3.11 km/s show no change in flux density except a minor flare at 56900 MJD.
82
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.41: Time series for selected velocity channels of the 1667 MHz RCP component in
the regular monitoring window. The features in the velocity channels 1.58 km/s, 1.62 km/s,
1.67 km/s, −0.92 km/s and −0.42 km/s show no variation. The two black + signs plotted on
the 1.58 km/s velocity channel show the times the flares are expected.
The 1667 MHz transition also showed variation in the behaviour of the time series in both
LCP and RCP as shown in Figure 2.40 and Figure 2.41 but the variation in the RCP component
was not significant until the end of the monitoring window. There was a general increase in
flux density over the monitoring window with flux density increasing from 16 Jy to 28 Jy in
the LCP component. Flares were also recorded at 56900 MJD and 57150 MJD. The LCP
component shows a dip in flux density just before each flare. Linear variation in flux density
barely above the noise over the period was recorded in the RCP component. For the methanol
line Figure 2.42 shows fifteen recurring flares with a periodicity of 246 days (Goedhart et al.,
2004) in the selected varying channels. The longer time range for the methanol data is because
83
 University of Ghana http://ugspace.ug.edu.gh
the data dates back from an earlier ongoing campaign. There was also a general increase in the
flux density. The variability in the velocity channel 1.2 km/s has been reported to be as intense
as 10% but stable for all other features (Caswell, Vaile and Ellingsen, 1995).
Figure 2.42: Time series for selected velocity channels of the 6668 MHz methanol maser tran-
sition in the regular monitoring window. The features at 1.18 km/s, 1.23 km/s and 1.27 km/s
show 15 periodic flares. Part of the data from 54900 MJD to 55600 MJD has been flagged
because of bad data. However the features in channel −0.97 km/s and channel −0.92 km/s
show a monotonic increase in flux from 50 Jy to 150 Jy.
2.3.10 Times series of some selected channels: intensive monitoring
The time series for the intensive monitoring window are presented in Figure 2.43 through Fig-
ure 2.47. The intensive monitoring window showed similar behaviour in variability for both
transitions of OH.
84
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.43: Time series for selected velocity channels of the 1665 MHz LCP component in
the intensive monitoring window. Features at 1.36 km/s, 1.45 km/s and 1.53 km/s show a
significant flare with a peak moderately variable in flux density from 18 Jy to 28 Jy. The
features in channels 3.64 km/s and 4.39 km/s show no variation. Some data from 57820 MJD
to 57890 MJD have been flagged because of RFI.
85
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.44: Time series for selected velocity channels of the 1665 MHz RCP component in
the intensive monitoring window. The features in channels 1.31 km/s, 1.71 km/s, 1.88 km/s
and 1.93 km/s show no variation in flux density. The data from 57820 MJD to 57890 MJD
have been flagged because of RFI.
86
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.45: Time series for selected velocity channels of the 1667 MHz LCP component in
the intensive monitoring window. The features at 1.62 km/s, 1.67 km/s and 1.75 km/s show
significant variation of the peak from 10 Jy to 30 Jy whilst the features at 2.19 km/s and
2.41 km/s show no variation.
From Figure 2.43 and Figure 2.45 the time series for the 1665 MHz and 1667 MHz lines
show short term variation of flux density for the LCP components. Variations within the noise
levels of the RCP components of both transitions are shown in Figure 2.44 and 2.46. For both
spectra of the 1665 MHz line the data between 57820 MJD and 57890 MJD have been flagged
because of RFI. The methanol line is also shown in 2.47 with a flare between 57780 MJD and
57860 MJD.
87
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.46: Time series for selected velocity channels of the 1667 MHz RCP component in
the intensive monitoring window. The features in channels 1.4 km/s, 1.45 km/s, 2.15 km/s
are only slightly variable with a slow change in amplitude whilst the features in −0.84 km/s
and −0.97 km/s are weak and show no variation.
88
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.47: Time series for selected velocity channels of the 6668 MHz methanol maser tran-
sition in the intensive monitoring window. The features in the velocity channels 1.18 km/s,
1.23 km/s, 1.27 km/s, 1.31 km/s and 1.36 km/s show variability with a flare between
57780 MJD and 57860 MJD.
In comparison with the time series for the regular monitoring window, the intensive mon-
itoring window shows an increase in the flux density of the LCP component for 1665 MHz
whereas the flux density for the LCP component for 1667 MHz remained the same. The flux
density decreased for the RCP components in both transitions of the maser species.
2.3.11 Regular monitoring of some selected channels in all transitions
Figure 2.48 shows the time series of the most varying features of the LCP and RCP components
of the OH as well as the methanol lines. Five adjacent most varying features at the peak of each
main line are averaged and plotted for each transition. This was done after inspecting the
89
 University of Ghana http://ugspace.ug.edu.gh
spectrum so as not to combine peaks from other spots or double peaked spectra. The plots are
restricted to the period common to all species; that is, between 56700 MJD and 57200 MJD. In
the plot for the regular monitoring window, the 1665 MHz LCP component shows variations
with a noticeable flare from 56800 MJD to 57080 MJD but the data at peak of the flare have
been flagged because of RFI. The 1665 MHz RCP component, however, shows no variation
in flux density. The 1667 MHz LCP component shows significant variability with two flares,
the first rising steeply from 56850 MJD and falling off gradually at 56900 MJD and the second
rising at 57080 MJD and falling off at 57200 MJD. The 1667 MHz RCP component shows no
variation in flux density.
90
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.48: Some selected channels of the 1665 MHz and 1667 MHz hydroxyl maser lines in
both polarizations together with the 6668 MHz methanol maser line during the regular moni-
toring window.
The 6668 MHz transition shows significant variation with two flares, one at 56850 MJD
and the other at 57080 MJD separated by 246 days as calculated by Goedhart et al. (2004) and
91
 University of Ghana http://ugspace.ug.edu.gh
confirmed by van der Walt et al. (2009); van der Walt (2011b). Visual examination of the data
of the flare activities in Figure 2.48 indicates that for both OH transitions the flare occurs for a
duration of 120 days with the same separation as that of the methanol line. However the OH
starts to rise steeply 20 days after the methanol has risen and starts to decay gradually 30 days
after the methanol has decayed and settled into the quiet phase.
2.3.12 Intensive monitoring of some selected channels in all transitions
In the plot for the intensive monitoring window displayed in Figure 2.49 the 1665 MHz LCP
component exhibits variations with a flare rising at 57790 MJD and decaying at 57880 MJD.
The 1665 MHz RCP component does not show any form of variability. In the 1667 MHz LCP
component, a flare rising slowly at 57790 MJD and subsequently decaying at 57880 MJD is
noted. The 1667 MHz RCP component however shows no significant variability in flux density.
The 6668 MHz line shows a significant variability with a steep rise at 57780 MJD followed by
a slow decay which eventually plateaus at 57880 MJD with flux density 2500 Jy as before the
rise.
92
 University of Ghana http://ugspace.ug.edu.gh
Figure 2.49: Some selected channels of the 1665 MHz and 1667 MHz hydroxyl lines in both
polarizations and the 6668 MHz methanol line during the intensive monitoring window.
93
 University of Ghana http://ugspace.ug.edu.gh
2.4 Discussion
The monitoring campaign has revealed the variation of the LCP and RCP components of the
mainline OH maser transitions associated with the 6.7 GHz methanol line in G9.62+0.20E. A
velocity range of 30 km/s was used to show all detected components in the averaged spectra
of each polarization. For both maser species the main maser lines were spread from 0 km/s
to 3 km/s and no velocity shifts were recorded for both the mainline and the different features
in the spectrum. The results of the maser velocities presented in the foregoing sections are
consistent with OH systemic velocity reported in Hofner et al. (1996). The similarity in the
velocity spread of the main lines suggests that though both OH and methanol maser molecules
have different systemic velocities they may emanate from the same cloud position (Bachiller
and Perez- Gutierrez, 1997).
The flaring behaviour of the time series was in different forms for both methanol and OH.
The LCP components of the 1665 MHz and 1667 MHz OH masers showed the same flare
pattern. They both showed flares which were preceded by dips in flux densities. This pat-
tern is consistent with those reported by Goedhart et al. (2019). The RCP component in both
transitions showed sporadic variation which was only within the noise.
The similarity in the individual flare profiles of the 1665 MHz and 1667 MHz LCP compo-
nents suggests that the characteristics of the mechanism underlying the flaring affects both OH
masers in the same way. Both show a dip in flux density before a steep rise and gentle decay
followed by a quiet phase after which they dip and slowly increase to the next flare. However in
comparison with methanol whose flares are as well consistent with those reported by Goedhart
et al. (2004), there is a time delay in the onset of the flare of the hydroxyl masers of about 30
days, suggesting a correlated variation between the OH and methanol masers. Following the
discussion in Goedhart et al. (2019) the time lags are likely due in part to the difference in the
distances between the methanol and the OH masers from the HII region associated with them.
This is evidenced by the distribution of the OH and methanol masers relative to the HII region
in Figure 1 of Sanna et al. (2015), from which it is clear that the OH masers are excited further
from the HII region than the methanol masers. Both methanol and OH masers flare within
the same time window separated by a quiescent level which remains unchanged between flares
94
 University of Ghana http://ugspace.ug.edu.gh
indicating that the source of the flare is intrinsic to the masers themselves.
The profile of the methanol line has a steep decay profile compared to the OH line. The
slower decay of the OH masers has been discussed by Goedhart et al. (2019) within the col-
liding wind binary model to be due to a lower electron density for the background part of the
HII region associated with it. This is consistent with the OH masers being further out in the
HII region that has a decreasing radial density distribution as is needed for the radio SED as is
discussed in the next chapter. Observing the separate polarizations of the 1.6 GHz OH main-
line masers also confirmed the observed behaviour seen in Goedhart et al. (2019) total intensity
monitoring. From the results it was found that the peak flux of the main feature in the LCP
components of the OH masers and the total intensity of the methanol masers show variations.
There was a correlation between the LCP’s of both OH transitions and the methanol masers
regarding the period of the flare. Time delays between the two species were recorded as the
epochs were taken simultaneously. The OH masers arises almost a month after the methanol
has risen. Also the nature of the variability profiles were different for both transitions. OH
masers exhibited a steep rise and slow fall whereas the methanol exhibited a steep rise as well
as fast fall. The nature of the variability has been explained within the context of the colliding
wind binary model. The steep rise of the maser depicting the pulse created by the ionising ra-
diation and the gentle decay profile being the recombination of the hydrogen plasma as it goes
into its equilibrium state. With the explanation of the decay profiles for both transitions it can
be assumed the source of the flaring is within the vicinity of the masers.
95
 University of Ghana http://ugspace.ug.edu.gh
 University of Ghana http://ugspace.ug.edu.gh
CHAPTER 3
HIGH RESOLUTION STUDIES OF THE 1.6 GHz HYDROXYL
MASER SPOTS AND THE L-BAND RADIO CONTINUUM IN
G9.62+0.20E
The HII region associated with the 1.6 GHz OH maser lines in G9.62+ 0.20E has been ob-
served at L-band but not at high resolution. Because of this, the variable and non-variable
1.6 GHz maser spots presented in the previous chapter have been plotted on a 43 GHz con-
tinuum to locate their positions (Sanna et al., 2015; Goedhart et al., 2019). This continuum,
measured at a frequency of 43 GHz, is not the actual radio emission associated with the OH
masers at 1.6 GHz. Knowledge on the position, size and flux density of the HII region will help
locate the positions of the variable and non-variable masers relative to it.
High resolution observations of the variable and the non-variable OH maser spots in the
1665 MHz and 1667 MHz left and right circular polarization as well as the L-band continuum
associated with them were made with e-MERLIN. This is to accurately determine the position
of the maser spots relative to the L-band continuum which will aid in the explanation of the
source of variability for the masers.
In this chapter a brief description of radio interferometry is given and the observation,
data reduction and results for the L-band continuum and its associated 1.6 GHz OH masers
in G9.62+0.20E are presented. Section 3.1 gives information on the e-MERLIN interferomet-
ric array followed by Section 3.2 which gives a brief introduction to interferometry. The details
of the observation of the data are given in Section 3.3. In Section 3.4 the interferometric data
reduction techniques employed in the calibration process are described. Section 3.5 gives the
results for the reduction and Section 3.6 discusses the results for the observation.
97
 University of Ghana http://ugspace.ug.edu.gh
3.1 The e-MERLIN interferometric array
The observations for the 1.6 GHz OH spectral line (narrow band) and continuum (wide band)
data in G9.62+0.20E were made with the e-MERLIN interferometric array, a United Kingdom
national facility operated by the University of Manchester at Jodrell Bank Observatory (JBO)
on behalf of the Particle Physics Astronomy Research Council (PPARC). The e-MERLIN array,
shown in Figure 3.1, is a radio interferometer whose correlator is based at the Jodrell Bank
Observatory. The array consists of two local stations (Lovell and Mark II) and five remotely
linked dishes (Cambridge, Defford, Knockin, Darnhall and Pickmere) spread across the United
Kingdom (Belles et al., 2015). Five of the radio antennae have an identical diameter of 25 m
and the remaining two, Lovell and Cambridge, have diameters of 76 m and 32 m respectively.
Shown in Figure 3.1 are the locations of the 7 antennas in an array configuration which yields 21
baselines ranging from 11 km to 217 km. Cambridge and Knockin, separated by a distance of
217 km provide the longest baseline at all bands and determine the array’s maximum resolution.
The shortest baseline is provided either by Mark II or Lovell (since they share the same
Figure 3.1: The e-MERLIN interferometric array with seven antennas distributed across the
UK, with baselines spanning 11 km to 217 km. Displayed here are the seven antennas, namely;
Lovell, Mark 2, Cambridge, Defford, Knockin, Darnhall and Pickmere (Belles et al., 2015).
98
 University of Ghana http://ugspace.ug.edu.gh
compound at JBO) and Pickmere located about 11.2 km from JBO. The array’s configuration
gives it an angular resolution comparable to the Hubble Space Telescope, making it capable
of carrying out centimetre-wavelength radio astronomy with micro-jansky sensitivities. Its
flexible wide-band correlator enables it to observe in three modes with three observing bands,
namely, L-band (1.23 GHz-1.74 GHz), C-band (4.3 GHz-7.5 GHz), and K-band (19 GHz–
25 GHz). At L-band an angular resolution of 160 mas is achievable whereas at C-band an
angular resolution of 50 mas can be achieved.
3.2 Interferometry
The interferometer makes measurements (observations) in the spatial frequency domain, sam-
pling the u-v coordinates in the plane perpendicular to the axis of an antenna. For a basic
two-telescope interferometer with antennas separated by a distance of bλ as shown in Figure
3.2, the incoming waves are converted into electric signals (voltages) and sent to a correlator
which multiplies and averages the signals to give the complex visibility V , defined by
∫
V (u,v) = I(l,m)e−2πi(ul+vm) dl dm (3.1)
The complex visibility V (u,v) = Aeiφ describes the amplitude A and phase φ of the signal at a
point in the u-v plane and thus contains wavelength information in addition to atmospheric and
instrumental terms. The signals add up to make an image of the sky, where u and v are defined
as the east-west and north-south positions with respect to the w axis parallel to S0 as shown
in Figure 3.3, and l and m are the direction cosines of S. The complex visibility therefore
is the Fourier transform of the brightness distribution I of the source from which the image
of the sky is obtained, each measurement of V (u,v) giving a single Fourier component of I
(Thompson et al., 2001). Hence, the intensity distribution can be recovered from the inverse
Fourier transform of the visibilities as
∫
I(l,m) = V (u,v)e2πi(ul+vm) dudv (3.2)
where 2π(ul + vm) is an approximate conversion required to change the observed phase φ =
99
 University of Ghana http://ugspace.ug.edu.gh
Figure 3.2: Basic telescope arrangement in an interferometer. Adopted from Burke and
Graham-Smith (2002).
Figure 3.3: Interferometery coordinates based on Burke and Graham-Smith (2002).
100
 University of Ghana http://ugspace.ug.edu.gh
φ(u,v) of V to the value that would be measured with antennas in the uv plane. The (l,m)
coordinates are the right ascension and declination, the actual celestial coordinates (position on
the sky) of the brightness I of the source that is used to determine the flux density.
3.2.1 Aperture synthesis and imaging
Aperture synthesis is employed to produce the imaging properties of an aperture by autocorre-
lating the radiation field at individual positions within the aperture. This technique uses a num-
ber of baselines from a configuration of telescopes to give multiple measurements of V (u,v)
for a given integration time. An array of n telescopes generate n(n− 1)/2 visibilities for an
individual integration. Thus for e-MERLIN operating without Lovell 15 baselines are avail-
able and the same number of visibilities are generated for each integration. For n telescopes
the earth’s rotation increases the number of projected baselines in the configuration with the
effect of filling the uv plane. The use of wider bandwidths also helps to fill the gaps in the
uv plane. The geometry of the telescopes in an interferometer in relation to the direction of
the source determines the instantaneous uv coverage of an interferometer. It is therefore the uv
coverage that determines in spatial frequency the scales of structure that can be measured by
the interferometer (Marr et al., 2016).
3.2.2 Calibration
To determine the brightness distribution of the source function, the visibility given by the
Fourier transform of the sky brightness is decomposed into components to find all the am-
plitudes and phases of the source. The correlator output is used for this purpose. The correlator
output V observedxy however differs from the output V
true
xy from an ideal interferometer by a factor
because of instrumental effects and propagation effects in the atmosphere and the ionosphere.
The goal of calibration is to correct the amplitudes and phases for these atmospheric and in-
strumental effects. The relationship between the observed visibilities and the true visibilities
on the baseline between the two telescopes x and y is expressed as
V observed = J truexy xyVxy (3.3)
101
 University of Ghana http://ugspace.ug.edu.gh
where Jxy represents all the errors affecting the the visibility measurement on the baseline xy.
Calibration consists of deriving the set of time and frequency dependent values of Jxy in order
to determine the best approximation of V true from V observed. This is done by using observations
of strong calibrator sources to determine the bandpass response of the phases and amplitudes.
Most of these calibrators are variable on time scales of weeks and months and their flux densi-
ties are not known accurately. However, some celestial objects are well characterised and are
easily modelled and their models may be used as primary calibrators. Corrections that make
an observation fit the calibrator models are determined by deriving changes in amplitude and
phase with respect to frequency and time. Thus the observation of the calibrators and their
models are used to derive the fluxes of the bandpass calibrator, flux density calibrator and gain
calibrator.
To set the flux density correctly a calibrator (flux density calibrator) with known spectral
shape and flux density is used. Gains (amplitude and phase) for the flux density calibrator and
the model are used to converts the counts to flux density in janskys. The calibrated gains of
the flux calibrator are compared with the gains of the other calibrators to estimate their flux
densities. After the flux scaling the bandpass calibration which solves the amplitude and phase
variations with respect to frequency follows. This is done by calculating the channel-to-channel
variations of the observed bandpass. A smooth function which matches the model to the data
is used to remove the outliers. To solve for the gain (amplitude and phase) variations with
time the integration-to-integration variations are determined and a smooth function is used to
remove the outliers. The solutions from the calibrations are then applied to the phase calibrator
and subsequently transferred onto the target by interpolation ready for imaging (Oozer et al.,
2010).
3.2.2.1 Imaging
Practically an interferometer is unable to fully sample the uv plane due to its nature. During
the observation the source is observed for some time, and as the earth and the telescopes on it
rotate with respect to the source the pattern of uv coverage also rotates allowing more parts of
the uv plane to be sampled (Oozer et al., 2010).
102
 University of Ghana http://ugspace.ug.edu.gh
Figure 3.4: The uv coverage of e-MERLIN for the target during the observation.
Figure 3.4 is the uv coverage of e-MERLIN for the target used in this thesis. The tracks
show the projected baselines in units of wavelength (Mλ ) in the uv plane. As can be seen, it
is impossible to fill the uv plane resulting in an insufficient number of Fourier components to
describe the source. To adequately map the source, a function S(u,v) which samples the uv
plane is introduced into Equation 3.2, which now becomes
∫
ID(l,m) = V (u,v)S(u,v)e(i2π(ul+vm)) dudv (3.4)
where ID(l,m), the dirty image or dirty map, is related to the intensity distribution I(l,m) by
∫
I(l,m) = ID(l,m)∗B(l,m)dudv (3.5)
Here, ∗ denotes the convolution operation and
103
 University of Ghana http://ugspace.ug.edu.gh
∫
B(l,m) = S(u,v)e(i2π(ul+vm)) dudv (3.6)
is the dirty beam, synthesised beam or point spread function. This dirty beam and dirty image
are determined from the visibility data.
To get the true sky brightness distribution the dirty beam is deconvolved from the dirty
image in an iterative process using a common deconvolution algorithm in radio-astronomy
called clean (Högbom, 1974). Clean considers a source to be a collection of point sources or
delta functions. The clean process identifies a surface brightness (peak flux) in the dirty image,
fits a dirty beam to this position and multiplies it by a gain factor. This gives a value which is a
fraction of the dirty beam. The fraction of the dirty beam that has been scaled to the brightness
peak is then subtracted from the collection of dirty beams from the uv data. Since the peak
flux and the dirty beam are both considered as point sources, clean has the effect of removing
all intensity components related to the peak, that is, side lobes and artefacts. A residual image
is left behind whose position and peak flux are stored as clean components. This process is
repeated until either the threshold noise is reached or there are no points of flux greater than
the specified noise level. The clean components are convolved with the synthesised beam and
added to the residual map to produce the final image. The synthesised beam is the full width at
half-maximum (FWHM) of the main lobe of the dirty beam (Mcmullin, 2009).
3.3 Observation
The total observing time of the data (the target and the calibrators) was 21.36 hours out of
which 10.32 hours was spent on source. The observation comprised datasets covering almost
four consecutive days. Table 3.1 gives the breakdown of the date and time allocated for each
session of observation. Each observing run consisted of about 75 scans made up of 60 minutes
on the flux calibrator 1331+305, 60 minutes on the bandpass calibrator 1407+284, a sequence
of 3 minutes on the phase calibrator 1751−1950 and 7 minutes on the target 1806−2031 which
is repeated for the total on-source duration, and finally 60 minutes on the polarization calibrator
0319+ 415. To account for any change in effect during the time of the observation the scans
104
 University of Ghana http://ugspace.ug.edu.gh
on the calibrators were scheduled before during and after the target scans.
Table 3.1: Data observation schedule and total on source time.
Observation ID Run Date Total time (hrs)
0 20170616_L 16 June 2017 2.10
1 20170617_L 17 June 2017 1.63
2 20170618_L 18 June 2017 4.20
3 20170619_L 19 June 2017 2.39
Different correlator set-ups were employed for the continuum and spectral line data though
the data were taken simultaneously. The continuum data were taken in full polarization over
a frequency range of (1.25 GHz-1.76 GHz) with a total bandwidth of 512 MHz which was
divided into 8 adjacent spectral windows each with 64 channels. An integration time of 1 s was
used and the data were later averaged. The high spectral resolution data spread across two spec-
tral windows were taken simultaneously with the continuum data. Two 125 kHz bands each
centred on either of the two rest frequencies 1.6654018 GHz and 1.6673590 GHz were each
overlaid with 512 spectral channels resulting in a high resolution channel width of 0.244 kHz.
Frequency switching was used during these observations, cycling between the 1.665 GHz and
1.667 GHz line frequencies to provide data on both transitions. Table 3.2 and Table 3.3 sum-
marise the observing parameters used for the continuum and the spectral line mode respectively.
Table 3.2: Observing parameters for the L-band continuum at a central frequency of 1.5 GHz.
Antenna labels: De: Defford, Cm: Cambridge, Da: Darnhall, Kn: Knockin, Mk2: Mark 2, Pi:
Pickmere.
Parameter Specification
Project CY5204
Band L
Antennas Mk2, Pi, Da, Kn, De, Cm
Number of sources 6
Integration time (s) 1.0
Frequency range (GHz) 1.25–1.76
Number of spectral windows 8
Channels per spectral window 128
Channel width (MHz) 0.5
Spectral window bandwidth (MHz) 64
Total bandwidth (MHz) 512
Polarizations RR, RL, LR, LL
105
 University of Ghana http://ugspace.ug.edu.gh
Table 3.3: Observing parameters for the 1.6 GHz OH maser lines observation.
Parameter Specification
Project CY5204
Band L
Integration time (s) 1.0
Central Freq 1.665 GHz 1.6654775 GHz
Central Freq 1.667 GHz 1.6674775 GHz
Num. spw 2
Channels/spw 512
Channel width (kHz) 0.244
Spw bandwidth (kHz) 125
Polarizations RR, RL, LR, LL
Table 3.4 also contains the observed calibrators and their scaled flux densities. The flux cali-
brator was used as a primary calibrator and the phase calibrator, bandpass calibrator and polar-
isation calibrator were used as secondary calibrators.
Table 3.4: Calibrators used for the 1.6 GHz OH observations and their flux densities.
Calibrator Flux density (Jy)
Flux density calibrator (1331+305) 13.8±0.2
Phase calibrator (1751−1950) 0.057±0.002
Band pass calibrator (1407+284) 1.013±0.003
Polarization, bandpass (0319+415) 16.41±0.06
The recorded signals were correlated at the JBO correlator.
3.4 Data reduction
The data were reduced with the e-MERLIN CASA pipeline (version v0.10.02). Further data
processing imaging and analysis were carried out using the Common Astronomy Software Ap-
plications (CASA) (version 5.4.2-1, (Mcmullin, 2009)) and IPython (version 5.1.0) packages
using standard procedures. The pipeline is optimised for the calibration of both continuum
and spectral line data. The pipeline was run twice on each data set for pre-processing and
calibration.
106
 University of Ghana http://ugspace.ug.edu.gh
3.4.1 Pre-processing data
During the first run the data were pre-processed by converting the visibilities in raw uv fits to a
measurement set (MS). Here the uv fits were adjusted for Earth rotation, the data were sorted,
and scan numbers were adjusted to increment with each source change. The continuum data
was then averaged over every 8 channels and 8 seconds.
The coordinates (data position) of the phase calibrator used for the observation had an offset
of half an arcsecond as compared with that from the astrogeo (catalogue) website ( VLBI Cal-
ibrator search); hence, the coordinate was corrected with the correct position information from
α h m s δ ◦ ′ ′′the website ( (J2000)= 17 51 41 .343700 (J2000)= −19 50 47 .50500). Correction of
the phase calibrator position was important because of the low elevation of the source. At low
elevations such as that of the source as seen in Figure 3.5 there are phase instabilities arising
in part from the longer path length through the atmosphere. The separation between the target
and the phase calibrator of 3.5 degrees also introduced phase transfer errors.
107
 University of Ghana http://ugspace.ug.edu.gh
Figure 3.5: Figure showing the elevation of the individual sources in the observation. Scans
of the calibrators are coded by colour. The first field, coloured green, is the flux calibrator
followed by the bandpass calibrator coloured black. The phase calibrator and the target are
coloured brown and orange respectively. And the last calibrator coloured pink is the polar-
isation calibrator. The phase calibrator and the target have low elevations because of their
positions on the sky.
The three data sets were concatenated into one measurement set. A new source G9.745+
00.106 in the field of view of the target was phase shifted to the center of the field to make
it an independent source in the concatenated measurement set. This was calibrated and used
as an independent source for the self calibration. The spectral line data were split from the
continuum data, separating them into two measurement sets.
Bad data were excluded from the data sets using a number of processes called flagging. The
automated flagging software AOflagger (Offringa et al., 2010) adapted to e-MERLIN, which
uses specific strategies for known RFI at specific frequencies, was used to flag the RFI in the
data. The software also used routines based on the deviation from expected spectral behaviour
for more flagging. An a priori routine based on knowledge from the e-MERLIN setup (for
example, antenna slewing time, flagged autocorrelations, the end channels (0∼ 3,508∼ 511),
and any other failing antennae) was also used to flag bad data. The initial 300 seconds of the
108
 University of Ghana http://ugspace.ug.edu.gh
scans of the bandpass, the flux and the polarization calibrators were flagged. For the target
and the phase calibrator the initial 35 seconds of each scan was flagged. From the automatic
flagging and manual flagging carried out on the data a total of about 40% of the data was
flagged, a fairly typical percentage for e-MERLIN L-band data. At this stage plots of the
sources were generated for inspection. An example is Figure 3.4 which shows the visibility
coverage of e-MERLIN for the target. This plot gives an indication of how elongated the
synthesised beam looks like.
3.4.2 Continuum calibration
In the second run, the data were plotted and manually inspected by going through all baselines,
spectral windows and channels of each calibrator and the target to see if any obviously bad
data needed flagging. Flag commands for bad data were written into an external input file.
This included end channels that were not flagged in the pre-processing run and bad data which
the automatic flagger failed to remove. The pipeline then used the external file for additional
flagging and calibrated the data. In the following sections solutions from 20170617 _L are used
as illustrations for the data calibration process. The solutions for the other data are similar to
20170617_L hence the decision to use it for the illustration.
3.4.2.1 Initial calibrations
Prior to calibrations, the delays of each antenna relative to the reference antenna Mk2, which
has many short baselines and is likely to give good solutions for all the calibrators, were cal-
culated. These antenna-based delays, which give values per spectral window per polarization,
puts a phase limit to the frequency and channels in each spectral window. Figure 3.6 shows the
delays per polarization across the bandpass for each antenna relative to the reference antenna
for all the calibrators for the observation. The delays with good scatter indicate stability in time
with good signal-to-noise ratio. Antenna 1 however, deviates by as much as 100 nanoseconds
as can be seen in Figure 3.6. Generally a deviation by this much is still a suitable solution since
the delays are stable in time.
For good time-dependent bandpass solutions the average over the small variations of phase
109
 University of Ghana http://ugspace.ug.edu.gh
110
Figure 3.6: Plot of the delays for the entire duration of an observation on all the calibrators. Each antenna’s delay relative to the reference antenna is
plotted in a separate pane with the blue and green colour codes showing the different polarization components. Here Antid=0 represents Darnhall
, Antid=1 represents Knockin, Antid=2 represents Defford, Antid=3 represents Mark2, Antid=4 represents Pickmere, and Antid=5 represents
Cambridge.
 University of Ghana http://ugspace.ug.edu.gh
with time in the bandpass with respect to the observing frequency is needed. This initial solu-
tion accounts for gain variations between different scans of the calibrators, especially if some
scans were taken at different elevations. Preliminary time-dependent estimates of the phase and
amplitude on all calibrators were thus calculated. These consisted of initial sets of antenna-
based gains with solution intervals of 16 seconds for phase and 120 seconds for amplitude on
all calibrators in order to moderate the effects of variations from integration to integration and
from scan to scan. The average phase variations with time help solve the de-correlation when
averaging for the final bandpass solution. The amplitude variations, however, have little effect
on the bandpass solutions.
Plots of the calibration solutions showing variations in the phases and the amplitudes of the
visibilities for all calibrators were generated. In Figure 3.7 the phase solutions scatter around
zero by a few degrees for the phase calibrator with each scan being coloured separately by their
spectral windows and a flat phase for the other calibrators (primary). Figure 3.8 shows the
amplitude solutions with some discrepancies in amplitudes when the calibrators are compared
to each other. The difference in amplitude shows a much weaker phase calibrator as compared
to the flux calibrator. For individual calibrators both phase and amplitude plots are flat showing
that averages in frequency can be made across a spectral window. Plots of this nature were
generated after each process and were used to inspect the quality of the calibrations. The
inspection is performed after each stage of the calibration. The calibrations were applied to
the data to produce flat phases as a function of frequency. These calibration solutions were
discarded after they had been applied. In the calibration plots that follow, Antid=0 represents
Darnhall, Antid=1 represents Knockin, Antid=2 represents Defford, Antid=3 represents Mark2,
Antid=4 represents Pickmere, and Antid=5 represents Cambridge. The calibration step was
followed by flux density scaling.
3.4.2.2 Flux density scaling
To effectively set the scale of flux density by which the target source is to be measured, the
absolute flux density scale was found by bootstrapping the flux densities of the calibrators to
the flux density of the flux calibrator 3C286 at L-band. This was done with a model derived
111
 University of Ghana http://ugspace.ug.edu.gh
112
Figure 3.7: Initial gain (phase) calibration solutions for all calibrators coloured by spectral windows. Here Antid=0 represents Darnhall, Antid=1
represents Knockin, Antid=2 represents Defford, Antid=3 represents Mark2, Antid=4 represents Pickmere , and Antid=5 represents Cambridge.
 University of Ghana http://ugspace.ug.edu.gh
113
Figure 3.8: Initial amplitude calibration solution for all calibrators. In the mosaic, the first field represents the polarisation solutions, the second
field represents the flux calibrator solutions, the third field represents the bandpass solutions, and the last field represents the scans of the phase
calibrator solutions. Here Antid=0 represents Darnhall, Antid=1 represents Knockin, Antid=2 represents Defford, Antid=3 represents Mark2,
Antid=4 represents Pickmere, and Antid=5 represents Cambridge.
 University of Ghana http://ugspace.ug.edu.gh
from a 12 hour observation of 3C286 on 5.5 GHz and scaled to the observing frequency in the
L-band, at 1.5 GHz. The flux spectral index and curvature were derived using model parameters
from Perley and Butler (2013). This excluded antenna Lovell since it did not partake in the data
collection.
Figure 3.9: Flux density scale for the calibrators at L-band observing frequencies. The blue
line shows the polarisation calibrator, the green line shows the phase calibrator and the red line
shows the bandpass calibrator. The upper panel shows the flux density and the spectral index
of each calibrator at the observing frequency.
The resultant values of flux density were then multiplied by the e-MERLIN factor of 0.9967
at L-band, calculated by Peck and Fenech (2013), to get a flux scale file which was used to
calibrate the appropriate fluxes of the calibrators. This step compensates for the loss in flux of
sources slightly resolved by e-MERLIN. Figure 3.9 is a plot showing the values of the scaled
flux densities of the calibrators. Also shown in the same plot is their spectral index information.
3.4.2.3 Gain calibration
Gain calibration commenced after setting the correct flux densities and spectral indices. Gain
calibration involved determining solutions of the amplitude gain and phase shift due to each
114
 University of Ghana http://ugspace.ug.edu.gh
antenna which together form the complex gain for that antenna. This was achieved in two steps
because phase and amplitude vary differently with time. The phase of the complex gain varies
frequently with time whereas amplitude and the bandpass response of the instrument across the
frequency band vary slowly over a long time. Obtaining these solutions in two parts deals with
both situations separately.
The first stage of the gain calibration is the amplitude calibration. This is derived by em-
ploying the measured antenna gain values to calculate the correction in voltage offsets in the
system. A solution interval of 120 seconds with the corrected flux densities of all calibrators
was used to calculate the amplitude gains against time. A plot of the amplitude calibration so-
lutions with scaled flux densities is shown in Figure 3.10. The solutions result from comparing
the different flux densities of all the calibrators except 1331+ 305 with a 1 Jy model. As can
be seen in Figure 3.10 the amplitudes are similar for all calibrator sources, spectral windows
and polarisation components.
In the next stage, which is phase calibration, the interference fringes calculated from two
antennas on each baseline are used to remove the errors in the phases introduced by the spectral
windows. This aims to produce flat phases with respect to the frequency (channel) in the data
by selecting a reference antenna for which the gain phase is defined to be zero. To achieve this
the group delay and phase rate against time for all calibrators using a short time interval of 8
seconds were calculated and applied to the whole data. Thus, the phase variations over the time
of the observation are removed.
Figure 3.11 shows the phase calibration solutions with scaled flux densities. The phase
solutions for each antenna as a function of time shows a similar time-dependence as for the
initial phase calibrations however the phases are not random. There are a few drifts in phase
but with no phase jumps.
3.4.2.4 Bandpass calibration
The last stage was the bandpass calibration which was done after the phase calibration step.
This corrects for the bandpass response across the observed bandwidth to compensate for any
frequency to channel differences in the system. A complex bandpass strategy which involved
115
 University of Ghana http://ugspace.ug.edu.gh
116
Figure 3.10: Amplitude gain solutions for all the calibrators, the scaled amplitudes for each antenna are plotted in a separate pane. Each of the
spectral windows is plotted with two polarization components shown by the different colours.
 University of Ghana http://ugspace.ug.edu.gh
117
Figure 3.11: Gain phase solutions of all the calibrators. Each of the spectral windows for each calibrator is plotted with two polarization components
shown by the different colours.
 University of Ghana http://ugspace.ug.edu.gh
a phase correction and an amplitude correction was employed to calculate the solutions on the
bandpass calibrator over the whole bandwidth at all times.
Figure 3.12 shows the complex phase solutions on the bandpass calibrator with flat phases
across the frequency bands. Small phase variations with frequency are a reflection of the
slightly different spectral windows across the bandpass.
In Figure 3.13, the complex amplitude solutions show a mean amplitude of approximately
1 Jy on all antennas across the useful parts of the band. The wiggles across the bandpass are
corrected. Polarisation calibration on the calibrators were not done during this work, this was
not much of a problem as OH masers are highly polarised and the percentage leakages across
both polarization are negligible.
3.4.2.5 Applying calibrations
At this stage the calibration solutions are applied to the target using the phase reference cal-
ibrator. Before that all the solutions of the delay calibrations, bandpass calibration and gain
calibrations (amplitude and phase) calculated with the scaled flux densities were applied to all
the calibrators one after the other. The measurement set has columns for the actual and cor-
rected data and a column for the models. The strategy is to specify the appropriate solutions to
be applied to the data column which will be written to the corrected data column. As the scans
of the phase calibrator are interleaved with the scans of the target, solutions of the averages
of the scans for the phase calibrator when linearly interpolated create a long stretch of scans
incorporating the scans of the target.
Figure 3.14 and Figure 3.15 are the phase and amplitude solutions of the scan averages on the
phase calibrator.
These solutions were then transferred to the target using linear interpolation. This is done be-
cause the phase calibrator is selected to be very close to the source. It is therefore assumed that
they both experience similar atmospheric conditions and hence, to a good approximation, suf-
ficiently good calibrations are transferred. Figure 3.16 shows the image of the phase calibrator
which recovers a flux of 53 mJy as shown in Table 3.5.
118
 University of Ghana http://ugspace.ug.edu.gh
119
Figure 3.12: Phase solutions of the bandpass calibration on the bandpass calibrator (1407+284). Each of the 8 spectral windows is plotted with
two polarization components shown by the different colours.
 University of Ghana http://ugspace.ug.edu.gh
120
Figure 3.13: Amplitude solutions of the bandpass calibration on the bandpass calibrator (1407+ 284). Each of the 8 spectral windows is plotted
with two polarization components shown by different colours. The edges of the bandpass are the fall off due to the instrumental demarcation of the
frequencies by spectral windows.
 University of Ghana http://ugspace.ug.edu.gh
121
Figure 3.14: Solutions of phase scan averages on the phase calibrator transferred onto the target. Each colour represent a spectral window and a
polarization component averaged over a scan.
 University of Ghana http://ugspace.ug.edu.gh
122
Figure 3.15: Solutions of amplitude scan averages on the phase calibrator transferred onto the target. Each colour represent a spectral window and
a polarization component averaged over a scan.
 University of Ghana http://ugspace.ug.edu.gh
Figure 3.16: Image of the phase calibrator with a deconvolved size of (622.61± 2.31) mas×
(129.9±0.10) mas at a position angle of (13.27±0.01) degrees. Shown in the lower left corner
of the image is the synthesized beam.
Table 3.5: Position and flux density of phase calibrator.
α Error δ Error Integrated flux density Peak flux density Rms
(J2000) (h m s) (′′) (J2000)(◦′ ′′) (′′) (mJy) (mJy/beam) (µJy/beam)
17 : 51 : 41.3437 0.0002 −19 : 50 : 47.5058 0.0009 53.08±0.09 52.84±0.22 58.93
3.4.3 Self calibration
Self calibration was employed to correct for the inaccuracies introduced by the transfer of solu-
tions from the phase calibrator to the target. This procedure essentially removed the effects of
the rapidly changing atmospheric conditions which had not been removed in the traditional
calibration. The flux density of the maser lines at L-band in the wide band data was not
bright enough to be used to self calibrate the continuum data hence the independent source
G9.745+0.106 was used. The phase center of the independent source was shifted to the phase
center of the target and was used for self calibration. Figure 3.17 is the L-band continuum
image of G9.745+0.106 in the field which recovers an integrated flux density of 91.5 mJy
123
 University of Ghana http://ugspace.ug.edu.gh
as given in Table 3.6. Previous 1.4 GHz observation of the source in a VLA survey reports a
recovered flux of 161 mJy (Becker et al., 1994).
Figure 3.17: An image of G9.745+0.106 in the field of view which was used for the self
calibration with a deconvolved size of (456±54) mas× (280±17) mas at a position angle of
(67±12) degrees. Shown in the lower left corner of the image is the synthesized beam of size
0.57 arcsecs×0.10 arcsecs at a position angle of 11.49 degrees.
Table 3.6: Position and flux density of G9.745+0.106.
α Error (′′) δ Error Integrated flux density Peak flux density Rms
(J2000) (h m s) (′′) (J2000)(◦′ ′′) (′′) (mJy) (mJy/beam) (µJy/beam)
18 : 06 : 50.2557 0.0082 −20 : 27 : 41.296 0.012 91.5±3.1 11.27±0.66 368.7
The model of this image was written to the model column of the measurement set. This
model image made after applying the phase calibration retains the absolute position informa-
tion. Starting from the point source model an iteration of a phase self calibration on G9.745+0.106
with solutions as presented in Figure 3.18 was calculated and these solutions were applied to
124
 University of Ghana http://ugspace.ug.edu.gh
all the 8 spectral windows. An iteration of amplitude and phase self calibration was calculated
using the model derived from the phase self calibration of G9.745+0.106. The results are
presented in Figure 3.19, which shows the amplitude and phase self calibration solutions. The
solutions from the self calibration were applied to all the 8 spectral windows in the continuum.
3.4.3.1 Imaging of the continuum at L-band
Total intensity maps of the continuum using line-free channels were made by cleaning the 8
spectral windows. The rms noise in the image was calculated by averaging the rms from five
different emission free areas of the image using boxes of the same size.
3.4.4 Spectral line calibration
The high spectral resolution data made up of two spectral windows in the measurement set that
was split from the continuum measurement set was used for the narrow band calibration. The
residual continuum in the spectral line data was subtracted from the spectral lines in the uv
plane by specifying the spectral line channels and fitting a first order polynomial to the spectral
line. The parameters used for the acquisition of the data are recorded in Table 3.3. The initial
40 seconds of each scan in the high resolution data were flagged.
3.4.4.1 Gain calibration
The phase offset between the wide and narrow band channelisation was calculated on the bright
calibrator 0319+ 415. This was done by importing the delay solutions in Figure 3.6 and gain
phase solutions in Figure 3.11 from the wide band calibrations and calculating the gain phase
solutions on the narrow band maintaining the same reference antenna. The calculation assumed
that the narrow-band spectral windows do not need delay corrections as any slope is negligible
over a narrow bandwidth. Usually there was no amplitude offset. The solutions of the offsets
are shown in Figure 3.20.
125
 University of Ghana http://ugspace.ug.edu.gh
126
Figure 3.18: Phase only solutions on G9.745+0.106 applied to the continuum with a spell of bad weather recorded between 20 : 20 hours and
21 : 30 hours. Some solutions were lost on Cambridge.
 University of Ghana http://ugspace.ug.edu.gh
127
Figure 3.19: Amplitude self calibration on G9.745+0.106 applied to the continuum with a spell of bad weather recorded between 20 : 20 hours
and 21 : 30 hours.
 University of Ghana http://ugspace.ug.edu.gh
128
Figure 3.20: Phase offset solutions between the narrow and wide band channelisation at 1665 MHz and 1667 MHz. The two polarisation compo-
nents are shown by the different colours.
 University of Ghana http://ugspace.ug.edu.gh
3.4.4.2 Bandpass calibration
The gain phase offset solutions were used to calculate the bandpass solutions on the narrow-
band bandpass. The same calibrator was used to derive the bandpass solutions in Figure 3.21
and Figure 3.22 exactly as for wide band but without delay corrections. This also assumed that
the phase and amplitude solutions derived from the wide band phase reference included the flux
scale hence no flux scaling was done.
The corrections in Figure 3.21 and Figure 3.22 were applied to the target on the narrow band
data. In this case the time-dependent offset corrections from the wide band spectral windows
were applied to the two spectral windows. The target was then split from the measurement set
and the measurement set was shifted to constant velocity of the hydroxyl maser lines.
3.4.4.3 Imaging of the spectral line
No further calibrations were carried out on the high resolution data. Spectral window 0 and
1 were set to the rest frequencies of 1.6654018 GHz and 1.6673590 GHz respectively. The
final image cubes of the left and right circular polarizations of each transition were produced
by identifying the maser emissions in each velocity plane of the cube and setting a clean mask
around them. The cubes were cleaned to the residual flux limit.
At this point, the data consisted of image cubes of 511 channels at 1024×1024 pixels per
Stokes parameter (LCP and RCP) per frequency. The velocity planes within which the masers
occurred were recorded. Components were considered as spectral features if they occurred in
three or more consecutive channels and had a peak flux density at least three times higher than
the rms noise in the field. The flux densities were measured by creating polygons around the
emissions in the image and recording the peak and integrated flux densities. A two dimensional
Gaussian component was fitted to the brightest peak in each channel map to obtain their po-
sitions and sizes. The uncertainties in the flux densities were calculated by averaging the rms
noise from five different emission-free areas of the image using boxes of the same size.
129
 University of Ghana http://ugspace.ug.edu.gh
130
Figure 3.21: Bandpass calibration on the narrow band data for the 1665 MHz maser line. The calculated solutions are coloured by the two
polarizations.
 University of Ghana http://ugspace.ug.edu.gh
131
Figure 3.22: Bandpass calibration on the narrow band data for the 1667 MHz maser line. The calculated solutions are coloured by the two
polarizations.
 University of Ghana http://ugspace.ug.edu.gh
3.5 Results
Presented in this section are the results from the imaging of the L-band continuum and the
maser emissions in the left and right circular polarizations of both the 1.665 GHz and 1.667 GHz
lines.
3.5.1 The L-band continuum
The continuum map was generated by imaging all the channels in the spectral bandwidth con-
taining emissions except the channels that contained the maser lines. Figure 3.23 shows the
site for the G9.62+ 0.20E star forming region with the expected position of the continuum
emission marked with a pink cross. At this position no radio continuum emission was recorded
above the e-MERLIN sensitivity limit. Measuring the rms noise gave a value of 54 µJy/beam
which is a 1σ limit on the peak flux density. At L-band with a target well above a declination
of 20 degrees, under good conditions and with vigorous RFI removal techniques a theoretical
noise limit of 14 µJy/beam is expected for a 12 hour continuum observation for e-MERLIN
without Lovell. However, for this work Table 3.1 reports a total on-source time of 10 hours.
When the e-MERLIN noise level is extrapolated to 10 hours a noise level of 15 µJy/beam is
expected but a noise level of 54 µJy/beam was recorded. This value which is almost four times
the expected value is likely due to the low elevation of the source as shown in Figure 3.5, bad
weather conditions as seen in Figure 3.18 and Figure 3.19, and also side lobes from nearby
sources G9.62+ 0.20C and G9.62+ 0.20D. The loss of the long north-south most sensitive
Cambridge baseline as seen in solutions from Figure 3.18 also resulted in the increase in the
noise. The upper limits on the flux density was obtained by finding the average of the abso-
lute values of the flux from five boxes taken around the pink cross. This resulted in a value of
54 µJy which gives a 3σ upper limit on the flux density of 162 µJy. The resulting upper limit
was significantly lower then the expected flux of 0.6 mJy which comes from extrapolating the
higher frequency measurements from Franco et al. (2000) using their spectral index of 0.95.
132
 University of Ghana http://ugspace.ug.edu.gh
Figure 3.23: Image of the field of the L-band radio continuum of G9.62+ 0.20E. The pink
cross shows the expected position of the radio continuum. Shown at the lower left corner of
the image is the synthesized beam of size 0.78 arcsecs× 0.57 arcsecs at a position angle of
−25.36 degrees.
The upper limit value when plotted on the spectral index plot of Figure 5 of Franco et al. (2000)
yields Figure 3.24. This plot will be discussed in the discussion section of Chapter 4.
133
 University of Ghana http://ugspace.ug.edu.gh
Figure 3.24: Spectral index plot of the G9.62+ 0.20E HII region. Shown in blue squares are
the flux density values from different high frequencies adapted from Franco et al. (2000). The
red line is the least squares fit to the higher frequency detections only. Plotted in green circle is
the 3σ upper limit on the flux density for the G9.62+0.20E star forming region at 1.5 GHz.
In the field of view of G9.62+ 0.20E two UCHII regions were detected. The northerly
source observed coincides with the position of G9.62+0.20C. Table 3.7 gives the position and
the flux density of G9.62+ 0.20C. A previous observation by Kurtz et al. (1994) at 8.6 GHz
measured a flux density of 175 mJy whereas at 4.9 GHz Garay et al. (1993) measured 50 mJy.
Figure 3.25 shows the continuum emission in G9.62+0.20C resolved into two objects. The flux
density quoted is for the two objects, the size of the structure however cannot be determined
by Gaussian fit hence an estimate of 3.93 arcsec× 2.31 arcsec is made by inspection. The
134
 University of Ghana http://ugspace.ug.edu.gh
measured flux of 22.1 mJy compares favourably with the flux of 26 mJy at 1.4 GHz measured
by Becker et al. (1994).
Figure 3.25: Image of the L-band radio continuum emission of G9.62+0.20C shown in colour
scale. Shown at the lower left corner of the image is the synthesized beam of size 0.78 arcsecs×
0.57 arcsecs at a position angle of −25.36 degrees.
The more southerly detection is the UCHII region associated with G9.62+ 0.20D. The
radio continuum emission from G9.62+0.20D is shown in Figure 3.26 and Table 3.7 presents
the positions and flux density, 11.84 mJy, recorded for the source. A flux density of 70 mJy at
4.9 GHz had been reported previously by (Garay et al., 1993) whereas a flux of 65.8 mJy was
measured at 8.6 GHz by Kurtz et al. (1994). At 1.4 GHz an upper limit of 17 mJy is reported
by Becker et al. (1994) but inspection of the cut out images indicates a source with a flux of
around 10 mJy which is consistent with the recorded flux of 11.84 mJy.
135
 University of Ghana http://ugspace.ug.edu.gh
Table 3.7: Positions and flux densities of the L-band continuum associated with source C, D
and E in G9.62+0.20. The expected position of the source E is adopted from Sanna et al.
(2015). For G9.62+0.20E the peak flux is given by the non-detection upper limit.
Source α Error δ Error Integrated flux Peak flux Rms
(J2000)(h m s) (′′) (J2000)(◦ ′ ′′) (′′) (mJy) (mJy/beam) (µJy/beam)
G9.62+0.20C 18 : 06 : 14.3522 0.010 −20 : 31 : 24.652 0.013 22.1±1.9 1.50±0.09 52
G9.62+0.20D 18 : 06 : 14.9514 0.011 −20 : 31 : 42.620 0.045 11.84±0.46 2.38±0.10 52
G9.62+0.20E 18 : 06 : 14.6562 − −20 : 31 : 31.671 − < 0.16 − 52
Figure 3.26: Image of the L-band radio continuum emission of G9.62 + 0.20D with a de-
convolved size of (2.090± 0.106) arcsecs× (0.781± 0.025) arcsecs and a position angle of
(1.9±1.4) degrees shown in colour scale. At the lower left corner of the image is the synthe-
sized beam of size 0.78 arcsecs×0.57 arcsecs at a position angle of −25.36 degrees.
When the continuum results for G9.62+0.20C in Figure 3.25 and G9.62+0.20D in Figure
3.26 are compared to the results for G9.745+0.106 in Figure 3.17 and the phase calibrator
in Figure 3.16 it can be seen that the beam size is larger and rounder than expected for the
full array. This is due to the loss of the long north-south Cambridge baselines as seen in the
solutions from Figure 3.18.
136
 University of Ghana http://ugspace.ug.edu.gh
3.5.2 Maser spots in the LCP and RCP components at 1.665 GHz and
1.667 GHz
From the emission channels in the cubes moment maps of the maser emissions were obtained.
Displayed from Figure 3.27 to Figure 3.37 are the moment maps showing the flux density
and velocity distribution of some of the LCP and RCP components of both the 1665 MHz
and 1667 MHz maser lines. The rms noise within the cube of both the 1665 MHz and the
1667 MHz maser spots were typically 200 mJy. Also presented in Table 3.8 are the positions
and flux densities of the maser spots. Some weak spots do not show in the maps because the
threshold range set to eliminate the side lobes in the image was above the flux density of these
spots. The spectral features here are compared with the results from Chapter 2.
Within the cube for the 1665 MHz maser lines two spots were found for the LCP component
with spots shown in Figure 3.27 and Figure 3.28. In Figure 3.29 the spectrum for the LCP
component of the 1665 MHz maser shows two lines at 1.64 km/s and 1.95 km/s. In comparison
these two velocity components appear blended in Figure 2.16 of Chapter 2.
137
 University of Ghana http://ugspace.ug.edu.gh
Figure 3.27: The flux density distribution for the 1.665 GHz LCP component over the intensity
range 7.5 Jy/beam to 100 Jy/beam. At the lower left corner of the image is the synthesized
beam.
138
 University of Ghana http://ugspace.ug.edu.gh
Figure 3.28: First moment map for the LCP component of the 1.665 GHz maser line shown in
colour scale. The maser spots in the 1.64 km/s and 1.95 km/s velocity channels are displayed.
At the lower left corner of the image is the synthesized beam.
Figure 3.29: The spectrum for the LCP component of the 1665 MHz OH maser line.
The RCP component of the 1665 MHz maser line had three spots of which only the spot at
velocity channel 1.38 km/s is shown in Figure 3.30 and Figure 3.31. The spectrum for the
139
 University of Ghana http://ugspace.ug.edu.gh
RCP component of the 1665 MHz maser line shown in Figure 3.32 however shows all three
features at velocity channels 1.38 km/s, 2.12 km/s and 5.29 km/s. The components of Figure
3.29 correspond with the spectral features in Figure 2.17 of Chapter 2.
Figure 3.30: The flux density distribution for the RCP component of the 1.665 GHz maser line
over the intensity range 2.5 Jy/beam to 100 Jy/beam. At the lower left corner of the image is
the synthesized beam.
140
 University of Ghana http://ugspace.ug.edu.gh
Figure 3.31: First moment map for the RCP component of the 1.665 GHz maser line shown in
colour scale. The maser spots in the 1.38 km/s velocity channel is displayed. At the lower left
corner of the image is the synthesized beam.
Figure 3.32: The spectrum for the RCP component of the 1665 MHz OH maser line showing
the maser lines at the velocity channels 1.38 km/s, 2.12 km/s and 5.29 km/s.
141
 University of Ghana http://ugspace.ug.edu.gh
Three spots were found in the cube of the LCP component of the 1.667 GHz maser line. Only
one maser spot at velocity channel 1.77 km/s is presented in Figure 3.33 and Figure 3.34. In
the spectrum for the LCP component of the 1.667 GHz maser line presented in Figure 3.35 all
three features at velocity channels −0.95 km/s, 1.77 km/s and 2.56 km/s are shown. These
features correspond with the spectral features in Figure 2.18 of Chapter 2 .
Figure 3.33: The flux density distribution for the LCP component of the 1.667 GHz maser line
over the intensity range 4.5 Jy/beam to 100 Jy. At the lower left corner of the image is the
synthesized beam.
142
 University of Ghana http://ugspace.ug.edu.gh
Figure 3.34: First moment map for the LCP component of the 1.667 GHz maser line shown in
colour scale. The maser spot in the 1.77 km/s velocity channel is displayed. At the lower left
corner of the image is the synthesized beam.
Figure 3.35: The spectrum for the LCP component of the 1667 MHz OH maser line showing
the spectral features at −0.95 km/s, 1.77 km/s and 2.56 km/s.
In the cube for the RCP component of the 1.667 GHz maser line four spots were recorded.
143
 University of Ghana http://ugspace.ug.edu.gh
Three maser spots in the velocity channels 0.63 km/s, 1.69 km/s and 2.21 km/s are shown
in Figure 3.36 and Figure 3.37. All four features are seen in the spectrum in Figure 3.38. In
comparison with Figure 2.19 of Chapter 2 all the spectral features are also seen.
Figure 3.36: The flux density distribution for the RCP component of the 1.667 GHz maser line
over the intensity range 1.3 Jy/beam to 100 Jy/beam. At the lower left corner of the image is
the synthesized beam.
144
 University of Ghana http://ugspace.ug.edu.gh
Figure 3.37: First moment map for the RCP component of the 1.667 GHz maser line shown in
colour scale. The maser spots in the 0.63 km/s, 1.69 km/s and 2.21 km/s velocity channels
are displayed. At the lower left corner of the image is the synthesized beam.
Figure 3.38: The spectrum for the RCP component of the 1667 MHz OH maser line showing
the spectral features at velocity channels −0.46 km/s, 0.63 km/s, 1.69 km/s and 2.21 km/s.
145
 University of Ghana http://ugspace.ug.edu.gh
More flux was measured in the LCP and RCP components of the 1665 MHz line than the
1667 MHz line. The flux measured in the LCP component of the 1665 MHz line was greater
than that measured for the RCP component. The 1667 MHz line followed this same pattern
but with lower flux densities in the RCP. The equatorial coordinate position measurements,
flux densities and velocity channels of the LCP and RCP components of the 1665 MHz and
1667 MHz maser are presented in Table 3.8. The first column of Table 3.8 gives the identity
number for the maser spots in this work. The second column gives the OH maser transitions
and their polarisations. In the third column the velocity channels within which the maser spots
are found are given. The fourth and the sixth columns specify the right ascension (α) and dec-
lination (δ ) of the maser spots determined by Gaussian fit. The leading terms of the positions
α ◦ ′are (J2000)= 18h 06m and δ (J2000)= −20 31 . The fifth and seventh columns give the cor-
responding errors for α and δ respectively. The eighth and ninth columns give the integrated
flux density and peak flux density respectively from the present work. In the tenth column the
rms noise levels in the image channels are given. The eleventh and the twelfth columns give the
maser feature numbers as they appear in the numbered rows of Table 2 of Sanna et al. (2015)
and Table of 3 Fish et al. (2005) respectively.
The flux densities measured in the 1.665 GHz and 1.667 GHz LCP and RCP components
are higher than the total intensity results in Figure 3 and Figure 4 of Goedhart et al. (2019).
Likewise, the peak flux densities listed in Table 3.8 of the present work are also higher than
those seen in the high resolution measurements of Sanna et al. (2015). For the LCP component
of the 1.665 GHz maser line spot ID 1 measured 16.08 Jy whereas its corresponding feature
12 in Sanna et al. (2015) measured 10.01 Jy. In the RCP component of the 1.665 GHz maser
line spot ID 3 measured 6.56 Jy whereas feature 18 in Sanna et al. (2015) measured 5.00 Jy. In
spot ID 4 a flux density of 3.10 Jy was measured whereas its corresponding feature in Sanna
et al. (2015) measured 4.35 Jy. For the LCP component of the 1.667 GHz line, the reference
and brightest maser spot in Sanna et al. (2015) referred to in the present work as spot ID 7
measured a flux of 22.97 Jy as against the 21.15 Jy reported in Sanna et al. (2015). Spot ID 6
measured 1.1 Jy against the 0.62 Jy reported in Sanna et al. (2015). The RCP component of
the 1.667 GHz as well followed the same pattern. Spot ID 9 measured 0.63 Jy more flux than
146
 University of Ghana http://ugspace.ug.edu.gh
147
Table 3.8: Absolute positions, velocities and flux densities of the polarisations of the mainline OH masers in G9.62+0.20E.
Maser spot ID Transition VLSR α Error δ Error Integrated flux density Peak flux density Rms Sanna feature Fish feature
(GHz) (km/s) (J2000)(s) (′′) (J2000)(′′) (′′) (Jy) (Jy/beam) (Jy/beam)
1 1.665 LCP 1.64 14.6788 0.0034 31.5694 0.0072 16.12±0.61 16.08±0.32 0.247 12 9
2 1.95 14.6901 0.0032 32.7603 0.0071 15.79±0.31 14.93±0.31 0.337 − 10
3 1.665 RCP 1.38 14.6402 0.0093 31.7291 0.009 8.33±0.40 6.56±0.19 0.115 18 3
4 2.12 14.6412 0.0085 31.749 0.013 3.46±0.24 3.10±0.12 0.109 19 1
5 5.29 14.6812 0.0094 32.738 0.018 1.32±0.07 1.10±0.13 0.187 − 10
6 1.667 LCP −0.95 14.6603 0.0068 31.922 0.012 1.98±0.13 1.79±0.06 0.153 3 −
7 1.77 14.6787 0.0028 31.5581 0.0057 22.97±0.68 22.04±0.35 0.107 1 25
8 2.56 14.6883 0.0076 32.786 0.016 2.57±0.21 2.32±0.10 0.103 − 26
9 1.667 RCP −0.46 14.657 0.016 31.970 0.029 1.21±0.11 0.89±0.17 0.101 11 −
10 0.63 14.644 0.012 31.763 0.030 2.17±0.38 1.96±0.24 0.157 8 21
11 1.69 14.674 0.012 31.597 0.020 6.05±0.51 4.74±0.23 0.190 7 25
12 2.21 14.6762 0.0071 31.566 0.015 3.60±0.28 3.42±0.14 0.139 9 24
 University of Ghana http://ugspace.ug.edu.gh
its corresponding feature 11 in Sanna et al. (2015). Spot ID 11 measured more than double the
flux density of its corresponding feature 7 in Sanna et al. (2015). The discrepancy seen when
the peak flux densities in the present work and the work of Sanna et al. (2015) are compared
arise in part from the variability of the masers and the higher resolution of the measurements
reported in Sanna et al. (2015).
On the positions of the maser spots, spot ID 7 which is the brightest corresponds to feature 1
in Table 2 of Sanna et al. (2015). This position is only off by 46 mas in RA and 16 mas in Dec.
The spot with ID 1 corresponds to feature 12 in Table 2 of Sanna et al. (2015) and feature 9 in
Table 3 of Fish et al. (2005). Maser spot ID 2 corresponds to feature 10 in Fish et al. (2005).
A cross reference of the positions of the spots with Sanna et al. (2015) and Fish et al. (2005)
are presented in Table 3.8. The relative positions of the remaining spots agree with Sanna et al.
(2015) within 50 mas. A comparison of the positional accuracy of 2 mas reported for the maser
spots in Sanna et al. (2015) and the offset of about 50 mas in the present work suggests that the
uncertainties in this work dominate.
3.6 Discussion
This chapter sought to find the positions of the 1.665 GHz and 1.667 GHz OH masers rel-
ative to their L-band HII region in G9.62+0.20E. This is the first L-band continuum study
for the G9.62+0.20E massive star forming region placing a constraint on the upper limit of
the detectable continuum flux density. The L-band continuum was not detected to a limit of
162 µJy. The non-detection could be attributed to the continuum being too weak to detect or
the continuum could be completely resolved out by the interferometer. However, at L-band the
size of the source as extrapolated from Q-band and C-band (see Chapter 4) is estimated to be
180 mas which is comparable to the e-MERLIN beam size of 150 mas. The estimated size of
180 mas in comparison with the Q-band size of 81 mas (Sanna, 2019) suggests the source is
getting larger at low frequencies. It is therefore likely the measured upper limit of 162 µJy on
the flux density is genuine and not an artefact, implying that the real flux density is less than
the predicted flux of 0.6 mJy. This seems to suggest that the actual spectrum turns over steeply
148
 University of Ghana http://ugspace.ug.edu.gh
at low frequencies. This will be discussed further in Chapter 4.
Twelve maser spots were detected from the LCP and RCP components of both 1.665 GHz
and 1.667 GHz in this thesis. The flux densities as determined by Gaussian fitting were typ-
ically similar to the results from Sanna et al. (2015). In comparing the absolute position of
the RCP and LCP components of both transitions of the 1.6 GHz OH masers with previous
results it has become apparent that both RA and Dec positions are in agreement with their cor-
responding peak positions in Table 2 of Sanna et al. (2015) to about 50 mas. The deviation
in the absolute positions can be accounted for by some factors: the position accuracy of the
phase calibrator and subsequent self calibration for the wide band data, and the declination of
the source. The positional accuracy of the target is inferred from the quality of the phase of the
phase calibrator. Self calibration on the extended G9.745+0.106 compromised the absolute
position information of the data. Even though no self calibration was done on the spectral line
results calibrations from the continuum which in itself had errors in addition to phase transfer
from the phase calibrator were used to calculate the offsets for the the high resolution band
calibrations. The transfer of the continuum information to the narrow band introduced uncer-
tainties into the narrow band data which could have affected the positions of the maser lines.
In addition if a brighter phase calibrator had been used there would have been no need for self
calibration of the continuum and the relative position of the continuum and maser would be
accurate even if the absolute position was compromised. For e-MERLIN greater accuracy in
position and sensitivity is achieved at higher declinations. Thus for G9.62+0.20E at a declina-
tion of −20 degrees the propagation of the radiation through the ionosphere compounded with
the uv coverage of the observation causes the beam to be very elliptical and elongated. This
means the low elevation of the source which causes the elongation of the beam in declination
presents a large error.
149
 University of Ghana http://ugspace.ug.edu.gh
 University of Ghana http://ugspace.ug.edu.gh
CHAPTER 4
HIGH RESOLUTION STUDIES OF THE 6.7 GHz METHANOL
MASER SPOTS AND THE C-BAND RADIO CONTINUUM IN
G9.62+0.20E
The periodically variable and nonvariable 6.7 GHz methanol maser spots in G9.62+ 0.20E
from Chapter 2 have been imaged simultaneously with the C-band continuum using the e-
MERLIN array. Presented in this chapter are the observation, calibration and results of the
observation of the 6.7 GHz masers and the C-band continuum.
Section 4.1 gives the observation details for the data acquisition. Aspects of the data reduc-
tion strategy employed for the 6.7 GHz maser lines and C-band continuum which differ from
the strategy employed for the 1.6 GHz line and L band continuum are given in Section 4.2.
Section 4.3 presents the results which are discussed and interpreted in Section 4.4.
4.1 Observation
The e-MERLIN interferometric array was employed to observe the 6.7 GHz methanol spectral
line (narrow band) and the C-band continuum (wide band) in G9.62+0.20E. A total of 60.689
hours were spent on the observations out of which 20.18 hours were on the source (science
target). Table 4.1 shows the date and time allocated for each session of the observations. During
the observations 75 scans made up of 60 minutes on the flux calibrator 1331+305, 57 minutes
on the bandpass calibrator 1407+284, a sequence of 3 minutes on the phase calibrator 1751−
1950 and 7 minutes on the target 1806− 2031, and finally 60 minutes on the polarization
0319+415 calibrator were made
151
 University of Ghana http://ugspace.ug.edu.gh
Table 4.1: Data observation schedule and total on source time.
Observation Id Run Date On source time (hrs)
0 20170624 _C 24 June 2017 4.20
1 20170711 _C 11 July 2017 4.08
2 20170714 _C 14 July 2017 4.20
3 20170716 _C 16 July 2017 3.50
4 20170717 _C 17 July 2017 4.20
All the data were taken in full polarization using a total bandwidth of 512 MHz. Four spectral
windows placed across the C-band frequency range 6.3 GHz-6.812 GHz were employed for
the continuum data collection. Table 4.2 is the summary of the parameters for the continuum
data observation.
Table 4.2: Observing parameters for the C-band continuum at a central frequency of 6.5 GHz.
Antenna labels: De: Defford, Cm: Cambridge, Da: Darnhall, Kn: Knockin, Mk2: Mark 2, Pi:
Pickmere.
Parameter Specification
Project CY5204
Band C
Antennas Mk2, Pi, Da, Kn, De, Cm
Number of sources 5
Integration time (s) 1.0
Frequency range (GHz) 6.300–6.812
Number of spectral windows 4
Channels per spectral windows 128
Channel width (MHz ) 1
Spectral window bandwidth (MHz ) 128
Total bandwidth (MHz ) 512
Polarizations RR, RL, LR, LL
The continuum and the spectral line data were taken simultaneously. For the spectral line
data acquisition an additional spectral window of spectral bandwidth 1 MHz was placed within
spectral window 2 of the continuum configuration centred at the expected spectral line central
frequency of 6.66866 GHz. The 1 MHz bandwidth spectral window was overlaid with 512
spectral channels. This resulted in the further channelisation of spectral window 2 of the con-
tinuum configuration into the high resolution channels. Table 4.3 gives the parameters used for
the acquisition of the spectral line data
152
 University of Ghana http://ugspace.ug.edu.gh
Table 4.3: Observing parameters for methanol lines at 6.7 GHz.
Parameter Specification
Project CY5204
Band C
Number of sources 5
Integration time (s) 1.0
Central frequency (GHz) 6.66866
Number of spectral windows 1
Channels per spectral windows 512
Channel width (MHz) 1.95
Spectral window bandwidth (MHz)) 1
Polarizations RR, RL, LR, LL
Presented in Table 4.4 are the calibrators used for the observations and their flux densities
at C-band. The flux calibrator was used as primary calibrator and the phase, bandpass and
polarisation calibrators were used as secondary calibrators.
Table 4.4: Calibrators used for the observations at 6.7 GHz and their flux densities.
Calibrator Flux density (Jy)
Flux calibrator (1331+305 ) 6.5±0.1
Phase calibrator (1751−1950) 0.0832±0.0003
Band pass calibrator (1408+284) 1.51±0.01
Polarization, bandpass (0319+415) 35.90±0.22
4.2 Data reduction
The data were calibrated with the e-MERLIN CASA pipeline (version v0.10.02). Further data
processing, imaging, and analysis were carried out using the Common Astronomy Software
Applications (CASA) (version 5.4.2-1, (Mcmullin, 2009)) and IPython (version 5.1.0) pack-
ages using standard procedures. The pipeline processed the data using two steps on each data
set, that is, the pre-processing and the calibration, for both continuum and spectral line.
4.2.1 Preprocessing data
For each of the data sets, preprocessing of data involved converting the fits to measurement sets
(MS) as discussed in the previous chapter. The position of the phase calibrator was corrected
as was done for the L-band data during the conversion to measurement set. The spectral line
153
 University of Ghana http://ugspace.ug.edu.gh
data were split from the continuum data into a separate measurement set since they were taken
simultaneously. The continuum data was then averaged after every 4 channels and 4 seconds.
An a priori flagging routine was employed to flag the beginning and end of the channels of
the data, and failing antennas. The beginning of each scan and any data that deviated from
expected spectral behaviour were flagged. The calibration solutions from 20170717_C are
used to illustrate the calibration processes in this chapter.
4.2.2 Continuum calibration
4.2.2.1 Data flagging
The data were manually inspected and the locations of the bad datasets were written into
an input file which was fed into the pipeline for flagging. For the observation made on day
20170711_C, antenna Knockin was flagged because of low sensitivity. The data were exam-
ined and the strongest maser line in spectral window 2 channel 112 on the target was masked to
prevent the automatic flagger which acts on bandpass calibrated data from flagging it. Manual
flagging of bad data which the automatic flagger failed to remove was also done.
4.2.2.2 Gain and bandpass calibration
The same standard gain and bandpass calibration procedures as described in Chapter 3 for L-
band were performed on the data after completing the flagging routines. However, during the
flux scaling the absolute flux density scale was found by setting the flux densities of all the
calibrators to that of 3C286 at C-band. This was multiplied by the e-MERLIN factor of 0.9393
to get a flux scale file which was used to calibrate the fluxes of the calibrators. The calibration
solutions with the scaled flux densities were applied to all the calibrators. Scan averages of the
phase and amplitude calibrations on the phase calibrator were plotted and are shown in Figure
4.1 and Figure 4.2 respectively. Figure 4.1 shows that the phase corrections were small as the
delay solutions were already applied. In Figure 4.2 the local deviations from the bandpass
shown in the amplitude solutions have been removed by assuming that the bandpass is flat in
each spectral window. The calibrations were then transferred from the phase calibrator onto the
target by linear interpolation. In the calibration plots that follows Antid=0 represents Darnhall,
154
 University of Ghana http://ugspace.ug.edu.gh
Antid=1 represents Knockin, Antid=2 represents Defford, Antid=3 represents Mark2, Antid=4
represents Pickmere, and Antid=5 represents Cambridge.
Figure 4.3 shows the image of the phase calibrator which recovers a flux of 74 mJy as
shown in Table 4.5.
Figure 4.3: Image of the phase calibrator with a deconvolved size of (234.59± 0.30) mas×
(27.391± 0.004) mas at a position angle of (7.721± 0.001) degrees. Shown in the lower left
corner of the image is the synthesized beam.
Table 4.5: Position and flux density of Phase calibrator.
α Error δ Error Integrated flux density Peak flux density Rms
(J2000) (h m s) (′′) (J2000)(◦′ ′′) (′′) (mJy) (mJy/beam) (µJy/beam)
17 : 51 : 41.34369 0.00002 −19 : 50 : 47.5058 0.0001 74.24±0.03 73.94±0.11 77.90
4.2.2.3 Self calibration on the continuum
The five measurement sets were concatenated into one measurement set after the mandatory
amplitude and phase calibrations. The target source was then split from the continuum data
into a separate measurement set to reduce imaging time. An image of the protected spectral
155
 University of Ghana http://ugspace.ug.edu.gh
156
Figure 4.1: Phase calibration solutions on the phase calibrator for day 20170717. Each scan consists of the spectral windows plotted with the two
polarization components shown by the different colours.
 University of Ghana http://ugspace.ug.edu.gh
157
Figure 4.2: Amplitude calibration solutions on the phase calibrator for day 20170717. Each scan consists of the spectral windows plotted with two
polarization components shown by the different colours.
 University of Ghana http://ugspace.ug.edu.gh
channel 112 was made and the model of the image was Fourier transformed and saved to the
model column of the measurement set. A phase self calibration of the spectral channel 112
was calculated and applied to all the four spectral windows of the continuum. An image of the
phase self calibrated line was made and the model was saved. A single iteration of amplitude
and phase self calibration of the spectral channel 112 was calculated and applied to the four
spectral windows. The continuum map was made by making a multi frequency spectrum image
of the four spectral windows excluding the maser line channel. These spectral channels also
avoided the 15 adjacent channels before and after the maser channels in the bandwidth
4.2.3 Spectral line calibration
The high resolution (spectral line) data in the measurement set that was split from the con-
tinuum measurement set was used for the narrow band calibration. The end channels and the
beginning of all scans in the data were flagged. All the data for baselines to antenna Knockin
on day 20170711_C were flagged as was done for the continuum data. The continuum data
was subtracted from the spectral lines in the uv plane by specifying the spectral line channels
and fitting a first order polynomial to the spectral line.
4.2.3.1 Gain calibration
For the gain solutions the phase offsets between the wide band and narrow band channelisation
on the bright calibrator 0319+ 415 were calculated. The gain calibration solutions from the
wide band was imported from the wide band calculations, usually with no amplitude offset.
This calculation assumed that the narrow band spectral windows needed no delay corrections
as any slope is negligible over a narrow bandwidth.
4.2.3.2 Bandpass calibration
The calculated phase offset between the wide and narrow band channelisation, and the delay
solutions from the wide band were used to calculate the bandpass solutions on the narrow band
bandpass exactly as was done for wide band. Presented in Figure 4.4 and Figure 4.5 are the
phase and amplitude calibration respectively.
158
 University of Ghana http://ugspace.ug.edu.gh
159
Figure 4.4: Phase component of the bandpass calibration on the narrow band data coloured by the spectral windows and the two polarization
components.
 University of Ghana http://ugspace.ug.edu.gh
160
Figure 4.5: Amplitude component of the bandpass calibration on the narrow band data coloured by the spectral windows and the two polarization
components.
 University of Ghana http://ugspace.ug.edu.gh
The phase and amplitude solutions derived from the wide band phase calibrator assumed
they included the flux scale. The offset calibrations were then applied to the target of the narrow
band data. The target was split from the measurement set and the measurement set was shifted
to the constant velocity of the methanol maser line. All corrections were then applied to the
target.
4.2.3.3 Self calibration on the maser line
An initial image of the maser line channel was made, Fourier transformed and saved. A phase
only self calibration of the maser line was calculated and the solutions were applied to the
whole high resolution data. Solutions of the phase only self calibration are presented in Figure
4.6. The model of the phase only self calibrated data was Fourier transformed and saved. An
amplitude and phase only self calibration of the image was calculated and applied to the data.
The amplitude solutions are shown in Figure 4.7.
A cube of all the channels was made and the channels with signals were cleaned. Images of
the maser spots were saved. The flux density and rms noise were acquired as was done on the
L-band data.
4.3 Results
Figure 4.8 shows the image of the C-band continuum. A Gaussian fit to the continuum gives
the position and flux density recorded in Table 4.6. The flux density of 2.25 mJy from the
free-free emission of the continuum in Figure 4.8 is consistent with the expected flux density
of 2.252 mJy as extrapolated from higher frequencies using a spectral index of 095 (Franco
et al., 2000) suggesting that most of the flux was recovered. The position however is offset by
320 mas in RA and 152 mas in Dec when compared with the Q-band JVLA-A position in Table
1 of (Sanna et al., 2015).
161
 University of Ghana http://ugspace.ug.edu.gh
162
Figure 4.6: Phase self calibration solutions on the narrow band data coloured by the two polarizations.
 University of Ghana http://ugspace.ug.edu.gh
163
Figure 4.7: Amplitude self calibration solutions on the narrow band data coloured by the two polarizations.
 University of Ghana http://ugspace.ug.edu.gh
Figure 4.8: Image of the C-band radio continuum of G9.62+0.20E shown in colour scale. At
the lower left corner of the image is the synthesized beam.
Measuring the rms noise gave a value of 50 µJy/beam. Under good conditions a theoret-
ical rms noise limit of 13 µJy/beam is expected for a 12 hour continuum observation at this
frequency. For a 20 hour observation a lower rms noise limit of 8 µJy/beam is expected how-
ever a rather high rms noise has been measured. As explained in the previous chapter this is
attributed to the low elevation of the source, bad weather conditions and also side lobes from
nearby source G9.62+ 0.20D. Figure 5 of Franco et al. (2000) shows the spectral index plot
of the flux density of the HII region associated with G9.62+ 0.20E at different wavelengths.
The index was estimated to be equivalent to the slope of the line of best fit of the plot of log-
arithm of flux density and frequency for G9.62+0.20E. At these frequencies a spectral index
of 0.95± 0.06 was calculated for this region(Franco et al., 2000). Flux densities at lower fre-
quencies are arrived at when this plot is extrapolated. Table 4.6 presents the continuum size of
G9.62+0.20E at Q-band and C-band.
At C-band the angular size of 130 mas converts to a physical size of 0.003 pc placing the source
164
 University of Ghana http://ugspace.ug.edu.gh
165
Table 4.6: Position and flux density of the C-band continuum of G9.62+0.20E and G9.62+0.20D. Also given are the beam dimensions and the
position angles of the maser spots.
Source α Error δ Error Integrated flux Peak flux Rms Deconvolved size Position angle
(J200)(h m s) (′′) (J200)(◦′ ′′) (′′) (mJy) (mJy/beam) (µJy/beam) Θma j.(mas)×Θmin.(mas) (◦)
G9.62+0.20E 18 : 06 : 14.6332 0.0022 −20 : 31 : 31.8232 0.009 2.25±0.20 0.520±0.001 50 165±42×97±23 176±28
G9.62+0.20D 18 : 06 : 14.9103 0.0077 −20 : 31 : 43.125 0.009 43.4±2.0 1.24±0.02 50 734±19×541±16 58.9±3.9
 University of Ghana http://ugspace.ug.edu.gh
in the hypercompact category in the Kurtz-Franco classification system (Kurtz and Franco,
2002). The VLA configuration A beam size is similar to the e-MERLIN C-band beam size;
therefore, the measured VLA Q-band size of 81 mas for this source (Sanna et al., 2015) trans-
lates to a physical size of 0.002 pc. The results presented in the present work suggest the source
is getting larger at low frequencies.
Figure 4.9: Spectral index plot of the G9.62+ 0.20E HII region. Shown in blue squares are
the flux density values from different frequencies adapted from Franco et al. (2000). The flux
density for the C-band continuum is shown with a green circle.
The C-band flux density of the HII region at 6.5 GHz has been added to the plot in Figure 5 of
Franco et al. (2000) and presented in Figure 4.9. Inclusion of the 6.5 GHz flux density result in
the spectral index calculation gives 0.94±0.05. The upper limit on the integrated flux density
from the expected position of the L-band continuum of G9.62+ 0.20E in Chapter 3 has also
166
 University of Ghana http://ugspace.ug.edu.gh
167
Table 4.7: Physical properties of G9.62+0.20E. The size of the Q-band continuum comes from Sanna (2019) based on the results in (Sanna et al.,
2015).
Band Beam size Position angle Convolved size Position angle Deconvolved size Position angle
(mas) (◦) Θma j.(mas)×Θ ◦ ◦min.(mas) ( ) Θma j.(mas)×Θmin.(mas) ( )
Q 93±50 −1.5 145±5×77±3 174±2 112±7×58±6 172±6
C 160±20 10.55 229±26×102±9 5±2 165±42×97±23 176±28
 University of Ghana http://ugspace.ug.edu.gh
been added to Figure 4.9 to produce Figure 4.10.
Figure 4.10: An updated spectral index plot of the flux density distribution of the G9.62+0.20E
region at different frequencies adapted from Franco et al. (2000). The update to the plot is the
flux density for the C-band continuum shown with a yellow circle and the 3σ upper limit of the
peak flux density at L-band also shown with a green circle.
Figure 4.10 shows the updated spectral index plot with the addition of the upper limit of the flux
density from the expected position of the continuum at L-band. The modified spectral index
for G9.62+0.20E is 0.94±0.06.
In the field of view of the C-band continuum in G9.62+0.20E is the source G9.62+0.20D
shown in Figure 4.11. Here G9.62+0.20C is resolved out. Presented in Table Table 4.6 is the
position and flux densities for source G9.62+0.20D.
168
 University of Ghana http://ugspace.ug.edu.gh
Figure 4.11: C-band radio continuum image of G9.62+ 0.20D shown in colour scale. At the
lower left corner of the image is the synthesized beam.
At 6.5 GHz the HII region is cometary in shape recovering a flux of 43 mJy whereas Kurtz
et al. (1994) recovers a flux of 65.8 mJy at 8.6 GHz. Becker et al. (1994) also reports a flux
of 36.6 mJy at 5 GHz. Thus at 6.5 GHz a flux of 50 mJy is expected. The recorded flux in
comparison with the expected flux suggests that most of the flux is recovered. The cometary
morphology is also consistent with the 8.6 GHz radio continuum image in Figure 1 of Testi
et al. (2000). In comparing the results of the positions in Table 4.6 with that in Table 3 of Kurtz
et al. (1994) there is an offset of 615 mas in RA and 105 mas in Dec for source D. This is
almost similar to the offset that source E’s continuum is off by in (Sanna et al., 2015). Thus
the offset for source D and source E are about 0.5 arcsecs to the west compared to the previous
positions (Kurtz et al., 1994; Sanna et al., 2015).
169
 University of Ghana http://ugspace.ug.edu.gh
4.3.0.1 Maser spots at 6.7 GHz
Only two bright maser spots with high signal-to-noise ratios in their velocity channels were
seen in the cube. Those with low signal-to-noise ratios were not seen and are not recorded
here. However of the two spots identified in the cube only one spot is shown in the maps
presented in Figure 4.12 and Figure 4.13. This is because the threshold range set to eliminate
the side lobes in the image was above the flux density of the weaker spot. The diagonal blue
feature in Figure 4.12 also seen as a red feature in Figure 4.13 is a side lobe
Figure 4.12: The flux density distribution for the 6.7 GHz methanol masers over the intensity
range 700 Jy/beam to 1500 Jy/beam. At the lower left corner of the image is the synthesized
beam.
The flux density for the bright maser spots is seen in Figure 4.12. The −0.26 km/s maser
spot is not seen because of the 700 Jy/beam threshold set to exclude the sidelobes of the
1.25 km/s spot but it is seen in addition to the 1.25 km/s spot in the spectrum in Figure 4.14.
170
 University of Ghana http://ugspace.ug.edu.gh
Figure 4.13: First moment map for the 6.7 GHz methanol masers shown in colour scale. The
maser spot in the 1.25 km/s velocity channel is displayed. At the lower left corner of the image
is the synthesized beam
Figure 4.14 presents the spectrum of the 6.7 GHz methanol maser lines showing the 1.25 km/s
line and the −0.26 km/s line. The components of Figure 4.14 correspond with the spectral
features in Figure 2.25 of chapter 2.
171
 University of Ghana http://ugspace.ug.edu.gh
Figure 4.14: The spectrum for the methanol masers at 6.7 GHz.
A Gaussian fitted to the maser spots gave the positions and flux densities presented in Table
4.8. The first column of Table 4.8 gives the velocity channels of the maser spots. The second
and fourth columns specify the right ascension (α) and declination (δ ) of the maser spots
determined by Gaussian fit. The leading terms of the positions are α (J2000)= 18h 06m and δ
◦ ′
(J2000)= −20 31 . The third and fifth columns give the error in α and δ respectively. The
sixth and seventh columns give the integrated flux density and peak flux density respectively.
The eighth column also gives the rms noise level in the image channels. The ninth and the tenth
columns give the beam dimensions and the position angles of the maser spots respectively. The
maser spot in the velocity channel 1.25 km/s corresponds to the LBA C-band results in velocity
channel 1.05 km/s of Table 1 in Sanna et al. (2015). The position is off by 9 mas in RA and
172 mas in Dec. For the −0.26 km/s spot when compared with feature 2 of the 6.7 GHz
spots in Table 2 of Sanna et al. (2015) an offset of 123 mas in RA and an offset of 67 mas
in Dec are recorded. The measured integrated flux density for both maser spots are however
consistent with the integrated flux in Table 1 and Table 2 of Sanna et al. (2015) for 1.25 km/s
and −0.26 km/s respectively.
172
 University of Ghana http://ugspace.ug.edu.gh
173
Table 4.8: Absolute positions, velocities and flux densities of the 6.7 GHz methanol maser spots in G9.62+0.20E.
VLSR α Error δ Error Integrated flux Peak flux Rms Deconvolved size Position angle
(km/s) (J2000)(s) (′′) (J2000)(′′) (′′) (Jy) (Jy/beam) (Jy/beam) Θ ◦ma j.(mas)×Θmin.(mas) ( )
1.25 14.6593 0.0004 31.776 0.013 3626.88±0.41 3237.34±0.17 43.04 551.8±18.3×104.7±7.9 182.6±1.6
−0.26 14.6587 0.024 31.794 0.090 653.28±0.33 99.95±8.46 6.48 611.02±27.37×164.48±0.97 158.45±0.73
 University of Ghana http://ugspace.ug.edu.gh
4.3.0.2 Map of maser spots on the C-band continuum
The positions of the maser spots are overlayed on the continuum in Figure 4.15. Here the
position of the continuum is shifted to the position reported by Sanna et al. (2015). This gives
an overview of the positions of the maser spots relative to the continuum.
Figure 4.15: The C-band radio continuum of the G9.62+0.20E with the positions of the maser
spots overlayed on it. The peak position of the 1.25 km/s maser spot is represented by a red
circle and the −0.26 km/s maser spot is marked by a yellow star. The peak position of the
1.05 km/s maser spot in Sanna et al. (2015) is shown with a black plus sign.
The maser spots are located on the edge of the continuum. The 1.05 km/s brightest maser spot
in Sanna et al. (2015) plotted on the continuum reveals that the 172 mas offset in Dec of the
brightest methanol spot is quite significant
4.4 Discussion
The HII region associated with G9.62+0.20E has been detected at 6.5 GHz. This is the first C-
band continuum detection of this source. The recovered flux density of 2.25 mJy is consistent
with the expected results of 2.252 mJy at this frequency. The size of the HII region places it
in the hypercompact HII region of the Kurtz-Franco classification system (Kurtz and Franco,
174
 University of Ghana http://ugspace.ug.edu.gh
2002). The shape of the HII region as revealed by the resolved morphology of the object is
elliptical which agrees with a similar elliptical shape and orientation found by Sanna et al.
(2015).
The spectral index plot of the flux density in Figure 5 of Franco et al. (2000) has been
updated with the addition of the C-band continuum and the upper limit of the noise level at
L-band to that in Figure 4.10. The spectral index which characterises the strength of the spectra
from the continuum has been modified from 0.95±0.06 to 0.94±0.06 for the G9.62+0.20E
star forming region. This suggests that there is a spectral index drop between C-band and
L-band confirming the turnover of the flux density at L-band. This index is consistent with
thermal emission from an optically thick HII region. As discussed by Marsh (1975) the turnover
of the radio spectrum at low frequencies can be explained using the truncated inverse-squared
density gradient distribution model. This radio continuum behaviour has been observed in the
spectra of some mass outflows (Marsh, 1975). Previously, Franco et al. (2000) invoked a steep
density gradient to explain the 0.95 spectral index but from the results of the C-band and L-
band continuum this model no longer explains the low frequency spectrum hence the Marsh
(1975) model is adopted instead. Thus the turnover of the spectrum at L-band can be explained
in terms of the truncated inverse-square density distribution model.
The turn over in the radio SED indicates that the source is going optically thick. This has
been checked by examining the brightness temperature as a function of frequency assuming
an electron temperature of 8000 K to estimate the optical depths. The peak flux at Q-band
corresponds to a peak brightness temperature of about 80 K translating to an optical depth of
0.01. The C-band peak flux puts the brightness temperature at about 4000 K translating to
an optical depth of about 0.5. These optical depth values show that at higher frequencies the
source is optically thin and remains optically thin at C-band. However, when the optical depth
is extrapolated to L-band it is then very optically thick.
The high resolution results of the masers from the e-MERLIN observation at a resolu-
tion of 40 mas have shown that the maser spots resolve into two spots with different veloci-
ties. However, with VLBA’s high resolution, the masers resolve into 52 spots (Sanna et al.,
2015). This shows that the maser observations in this work suffered from poor resolution
175
 University of Ghana http://ugspace.ug.edu.gh
and spatial blending of spots as compared to that reported by Sanna et al. (2015). The flux
density of the two maser spots 1.25 km/s and −0.26 km/s found in the high resolution ob-
servations were also found to be consistent with the results in Chapter 2 for velocity channel
1.23 km/s and−0.22 km/s and also with Goedhart et al. (2004)’s velocity channels 1.26 km/s
and −0.14 km/s. The nonvarying masers have not been detected. Figure 4.15 graphically
shows that the variable spot is on the edge of the actual C-band continuum that supplies the
seed photons as opposed to Fig 1 in Sanna et al. (2015) which was overlaid on the Q-band
continuum.
176
 University of Ghana http://ugspace.ug.edu.gh
 University of Ghana http://ugspace.ug.edu.gh
CHAPTER 5
CONCLUSION AND FUTURE RESEARCH
In this work the left and right circular polarization components of the 1.6 GHz mainline hy-
droxyl masers and the 6.7 GHz methanol masers have been used to study the massive star form-
ing region G9.62+0.20 E. This is the first time the LCP and RCP components of the mainline
OH masers in G9.62+0.20 E have been monitored. Monitoring of G9.62+0.20 E from 2011
to 2017 has shown the continual variability of the flux density of the 6.7 GHz methanol masers
and the left and right circular polarization components of the 1.6 GHz OH mainline masers.
The variability profile for the separate polarizations of the 1.6 GHz OH mainline masers also
confirmed the observed behaviour seen in Goedhart et al. (2019)’s total intensity monitoring.
Between the OH and the methanol profiles, the OH masers exhibited a dip before a steep rise
and a slow fall whereas the methanol exhibited a steep rise as well as a fast fall. Time delays
between the two species were recorded.
High resolution observation of the L-band and C-band radio continuum were conducted
with the e-MERLIN array. The results showed a radio continuum at C-band with a flux density
of 2.25 mJy consistent with the calculated flux density of 2.252 mJy and an angular size of
130 mas. For the L-band continuum an upper limit of 162 µJy on the flux density has been
calculated by finding the average of the absolute value of the flux density from five regions at
the expected position of the continuum. The continuum emission results from both the C-band
and the L-band suggest that the actual spectrum turns over steeply from C-band to L-band.
A truncated inverse-square density distribution model (Marsh, 1975) was more successful at
explaining the results than a steep density gradient model for the 0.95 spectral index Kurtz
and Franco (2002). Estimation of the optical depth from the peak brightness temperature of
178
 University of Ghana http://ugspace.ug.edu.gh
both L-band and C-band revealed that the continuum at C-band is optically thin whereas the
continuum at L-band is optically thick. The positions of detected masers in the high resolution
data at C-band have been plotted relative to the associated continuum. The results of the maser
spots on the continuum observations show that the bright periodic methanol spots are on the
edge of the observed C-band continuum consistent with the colliding wind binary model.
The work gives new evidence to support the correlated variability of circularly polarised
components of the 1.6 GHz OH mainline masers and the 6.7 GHz methanol maser in G9.62+
0.20E. The results of the radio SED constrain the density distribution of the actual frequen-
cies of the seed photons of the C-band and L-band continuum. This adds on to the wealth of
constraints from various observations and theoretical calculations put forward within the con-
text of the colliding wind binary model to explain the observed periodicity for the masers in
G9.62+0.20E. The results of these independent constraints such as the direct imaging of the
HII region provides direct constraints on the density distribution which presents a good chance
to test the model and predict the electron density distribution indirectly. With these results a
photo-ionisation nebulae model such as CLOUDY (van den Heever et al., 2019) which uses a
constant density can be modified by taking into account the density gradient of the HII region
and in particular the truncated inverse-square density distribution of the L-band implied by the
radio SED found in this thesis. This will take inputs such as the luminosity and density gradient
of the HII region to predict the size of the radio continuum (extent of ionisation) and also test
the variation of the radiation from the colliding wind binary model.
The LCP and RCP components of the 1.6 GHz OH maser saw a dip just before the flares
and also recorded time delays between the onset of their flares and those of the methanol maser.
The radio SED measured has also put constraints on the density distribution of the seed photons
in the HII region. In the context of the colliding wind binary model the electron density at the
background location of the HII region determines the decay time. Slower decays observed for
the 1.6 GHz OH maser imply a low electron density and VLBI imaging (Sanna et al., 2015)
showing OH masers farther away than methanol masers from the HII region which is consistent
with a density gradient. The above described photo-ionisation nebulae model could be fit with
the Marsh model (Marsh, 1975) used to explain the L-band spectrum behaviour to test if the
179
 University of Ghana http://ugspace.ug.edu.gh
variation of the radiation of the HII region is associated with the time delays and dips observed
in the 1.6 GHz OH maser. But will such a model be consistent with the decay time? Would the
model predict a dip in the L-band continuum seed photons in order to explain the 1.6 GHz OH
variability profile? If so, is this because of the optical depth or due to the penetration of hard
xrays as explained by (Goedhart et al., 2019)?
5.1 Improving observational constraints
The observation of the G9.62+0.20E region made with e-MERLIN interferometric data can be
improved by adopting different strategies. For instance, the astrometry of the methanol maser
spots and the C-band continuum did not work well to achieve the original aim of the thesis.
This has been attributed to a number of factors. First is the selection of a weak phase calibrator
which was too weak to give good solutions. A solution will be the use of a brighter source
with high position accuracy or the use of two phase calibrators one with strong flux density and
the other with good astrometry. Second was the adoption of self calibration which ultimately
sacrifices the astrometry information. There would have been no need for a self calibration if
the previous solutions were adhered to. The third factor was the loss of data at L-band due to
bad data and the high noise level in the data. At L-band it is almost impossible to get rid of all
the bad data. An attempt at this will leave insufficient data for calibration. With the addition of
one or more 30 m antennas at Goonhilly into the existing e-MERLIN radio interferometer in
future the the longer north-south baselines that will then be available will improve the resolution
in the declination direction giving a rounder beam shape (Heywood et al., 2011). In this way
the initial goal of the work will be achieved.
Further observations of the polarised 1.6 GHz OH and the 6.7 GHz methanol masers taken
simultaneously with the continuum emission from the HII region associated with them in
G9.62+0.20 E will be made with the VLA in the A configuration. A detection of the L-band
continuum will allow a better model to be developed.
Also other future work will be based on selecting from the 6.7 GHz methanol sample pro-
vided by Goedhart et al. (2004) and from the many more recent periodic sources discovered the
180
 University of Ghana http://ugspace.ug.edu.gh
sources that show similar variability profile as that of G9.62+0.20 E and conducting monitor-
ing observation of their 1.6 GHz OH maser polarization in search of more dips and time lags.
A follow up with high angular resolution observation of the flaring and non-flaring phases of
the masers and their associated continuum to ascertain if dips only occur if the continuum is
optically thick. These results will then be used to test the colliding wind binary model.
181
 University of Ghana http://ugspace.ug.edu.gh
REFERENCES
Abbott, B., Abbott, R., Adhikari, R., Ajith, P., Allen, B., Allen, G., Amin, R., Anderson, S.,
Anderson, W. and M., A. (2009a), ‘All sky LIGO search for periodic gravitational waves in
the early fifth-science-run data’, Phys. Rev. Lett. 102(11), 111102.
Abbott, B., Abbott, R., Adhikari, R., Ajith, P., Allen, B., Allen, G., Amin, R., Anderson, S.,
Anderson, W. and M., A. (2009b), ‘LIGO: The Laser Interferometer Gravitational-Wave
Observatory’, Rep. Prog. Phys. 72(7), 076901.
Abbott, B., Abbott, R., Adhikari, R., Ajith, P., Allen, B., Allen, G., Amin, R., Anderson, S.,
Anderson, W. and M., A. (2017), ‘LIGO:Observation of Gravitational Waves from a Binary
Neutron Star Inspiral’, Phys. Rev. Lett. 119, 16–20.
Adams, F. C., Lada, C. J. and Shu, F. H. (1987), ‘Spectral evolution of young stellar objects’,
ApJ 312, 788.
Andre, P., Thompson, D. W. and Barsony, M. (1993), ‘Submillimeter continuum observations
of rho Ophiuchi A - The candidate protostar VLA 1623 and prestellar clumps’, ApJ 406, 122–
141.
Andrews, J., Calzetti, D., Chandar, R., Elmegreen, B., Kennicutt, R., Kim, H., Krumholz, M.,
Lee, J., Mcelwee, S., O’Connell, R. and Whitmore, B. (2014), ‘Big fish in small ponds:
Massive stars in the low-mass clusters of M83’, ApJ 793, 4.
Araya, E. D., Hofner, P., Goss, W. M., Kurtz, S., Richards, A. M. S., Linz, H., Olmi, L. and
Sewitlo, M. (2010), ‘Quasi-periodic formaldehyde maser flares in the massive protostellar
object IRAS 18566+0408’, ApJL 717, L133–L137.
Argon, A. L. and Reid, M. J. (2003), ‘A class of interstellar oh masers associated with proto-
stellar outflows’, ApJ 593, 925–930.
Artymowicz, P. and Stephen , H. L. (1996), ‘Mass flow through gaps in circumbinary disks’,
ApJ 467, L77–L80.
Arzoumanian, D., Andre, P., Didelon, P., Konyves, V., Schneider, N. and Men’shchikov, A.
(2011), ‘Characterizing interstellar filaments with Herschel in IC 5146’, A&AS 529, L6.
Bachiller, R. and Perez- Gutierrez, M. (1997), ‘Shock chemistry in the young bipolar outflow
L1157’, ApJ 487, L93–L96.
Batrla, W., Matthew, H., Menten, K. and Walmsley, C. (1987), ‘Detection of strong methanol
maser towards galactic HII regions’, Nature 326, 49.
Becker, R. H., White, L. R., Helfand, D. and Zoonematkermani, S. (1994), ‘A 5 GHz VLA
survey of the galactic plane’, ApJS 91, 347–387.
Belles, P. E., Beswick, R., Argo, M., Jackson, N., Muxlow, T. and Richards, A. (2015), e-
MERLIN CookBook, The University of Manchester, Manchester.
182
 University of Ghana http://ugspace.ug.edu.gh
Beuther, H., Churchwell, E., McKee, C. and Tan, J. (2007), The formation of massive stars, in
B. K. K. and J. David, eds, ‘Protostars and Planets V’, Vol. V, University of Arizona Press,
Tucson, Arizona, p. 165–180.
Beuther, H., Walsh, A., Johnston, K., Henning, T., Kuiper, R., Longmore, S. and Walmsley, C.
(2017), ‘Fragmentation and disk formation in high-mass star formation: The ALMA view of
G351.77-0.54 at 0.06” resolution’, A&A 603, A10.
Beuther, H., Zhang, Q. and Hunter, T., Sridharan, T. and Bergin, E. (2007b), ‘The 105 L
high-mass protostellar object IRAS 23151+5912’, A&A 473, 493–500.
Billington, S., Urquhart, J., Figura, C., Eden, D. and Moore, T. (2018), ‘The RMS sur-
vey: Ammonia mapping of the environment of young mas- sive stellar objects II’, MNRAS
483, 3146–3167.
Bonnell, I. A. and Bate, M. R. (2006), ‘Star formation through gravitational collapse and com-
petitive accretion’, MNRAS 370, 488–494.
Bonnell, I. A., Bate, M. R. and Clarke, C. J.and Pringle, J. E. (1997), ‘Accretion and the stellar
mass spectrum in small clusters’, MNRAS 285, 201–208.
Bonnell, I. A., Bate, M. R. and Vine, S. G. (2003), ‘The hierarchical formation of a stellar
cluster’, MNRAS 343, 413–418.
Bonnell, I. A. and Davies, M. B. (1998), ‘Mass segregation in young stellar clusters’, MNRAS
295, 691–698.
Bonnell, I. A., Vine, S. G. and Bate, M. R. (2004), ‘Massive star formation: nurture, not nature’,
MNRAS 349, 735–741.
Breen, S. L., Ellingsen, S. P., Caswell , J. L. and Lewis, B. E. (2010), ‘12.2-GHz methanol
masers towards 1.2-mm dust clumps: quantifying high-mass star formation evolutionary
schemes’, MNRAS 410, 2219–2244.
Burke, B. F. and Graham-Smith, F. (2002), An Introduction to Radio Astronomy, 2nd edn,
Cambridge University Press, Cambridge.
Caselli, P. and Ceccarelli, C. (2012), ‘Our astrochemical heritage’, Astron. Astrophys. Rev.
20, 56.
Caswell, J. L. (1998), ‘Positions of hydroxyl masers at 1665 and 1667 MHz’, MNRAS 297, 2l5–
235.
Caswell, J. L., Green, J. A. and Phillips, C. J. (2013), ‘Parkes full polarization spectra of OH
masers – I. galactic longitudes 350 through the galactic centre to 41’, MNRAS 431, 1180–
1219.
Caswell, J., Vaile, R. and Ellingsen, S. (1995), ‘Variability of methanol masers’, Proc. Astron.
Soc. of Aust. 12, 37.
Caswell, J., Vaile, R. and Forster, J. (1995), ‘Methanol and hydroxyl maser positions’, MNRAS
277, 210–216.
Cesaroni, R., Galli, D., Lodato, G., Walmsley, C. and Zhang, Q. (2006), Disks around young O-
B (proto)stars:observations and theory, in B. K. K. and J. David, eds, ‘Protostars and Planets
IV’, Vol. IV, University of Arizona Press, Tucson, Arizona.
183
 University of Ghana http://ugspace.ug.edu.gh
Chen, H. V., Keto, E., Zhang, Q., Sridharan, T. K., Liu, S. and Su, Y. (2016), ‘A hot and massive
accretion disk around high mass protostar IRAS20126+ 4104’, ApJ 823, 125.
Clegg, A. W. and Cordes, J. M. (1991), ‘Variability of interstellar hydroxyl masers’, ApJ
374, 150.
Cohen, R. J. (1989), ‘Compact maser sources’, Rep. Prog.Phys. 52, 881–943.
Commerçon, B., Hennebelle, P. and Henning, T. (2011), ‘Collapse of massive magnetized
dense cores using radiation magnetohydrodynamics:Early fragmentation inhibition’, ApJ
L9, 742.
Condon, J. J. and Ransom, S. M. (2016), Essential Radio Astronomy, Princeton University
press, Princeton, New Jersey.
Cragg, D., Sobolev, A. M. and Godfrey, P. (2002), ‘Modelling methanol and hydroxyl masers
in star forming regions’, MNRAS 331, 521–536.
Cragg, D., Sobolev.A.M. and Godfrey, P. (2005), ‘Models of class II methanol masers based
on improved molecular data’, MNRAS 360, 533–546.
Das, A., Acharya, K., Chakrabarti, S. and Chakrabarti, S. (2008), ‘Formation of water and
methanol in star forming molecular clouds’, A&A 486, 209.
Davies, B.and Hoare, M. L. S., Hosokawa, T.and Oudmaijer, R., Urquhart, J., Mottram, J. and
Stead, J. (2011), ‘The Red MSX source survey: critical tests of accretion models for the
formation of massive stars’, MNRAS 416, 972–990.
De Buizer, J. (2000), A mid infrared imaging survey of star forming regions containing
methanol and water maser emissions, PhD thesis, Graduate school of the university of
Florida,Florida,USA.
De Buizer, J. M. (2003), ‘Testing the circumstellar disc hypothesis: a search for H2 outflow sig-
natures from massive young stellar objects with linearly distributed methanol maser’, MN-
RAS 341(1), 277–298.
De Buizer, J. M., Osorio, M. and N., C. (2005), ‘Observations and modeling of the 2-25 µm
emission from high mass protostellar object candidates’, ApJ 635, 452–465.
De Buizer, J. and Minier, V. (2005), ‘Investigating the nature of the dust emission around
massive protostar NGC 7538 IRS 1: Circumstellar disk and outflow?’, ApJL L151, 628.
Dobbs, C. L., Bonnell, I. A. and Clark, P. C. (2005), ‘Centrally condensed turbulent cores:
massive stars or fragmentation?’, MNRAS 360, 2–8.
Dunham, M. M. and Vorobyov, E. I. (2012), ‘Resolving the luminosity problem in low mass
star formation’, AJ 747, 52.
Dzib, S., Loinard, L. and Rodriguez, L. F. (2011), ‘VLBA Determination of the distance to
nearby starforming regions. vi. The distance to the young stellar object HW 9 in Cepheus
A’, ApJ 733, 71.
Eden, D. J., Liu, T., Kim, K.-T., Juvela, M., Liu, S.-Y., Tatematsu, K., Francesco, J. D., Wang,
K., Wu, Y., Thompson, M. A., Fuller, G. A., Li, D. and et. al. (2019), ‘SCOPE: SCUBA-
2 Continuum observations of pre-protostellar evolution – survey description and compact
source catalogue’, MNRAS 485(2), 2895–2908.
184
 University of Ghana http://ugspace.ug.edu.gh
Edris, K. A., Fuller, G. A. and Cohen, R. J. (2007), ‘A survey of OH masers towards high mass
protostellar objects’, A&A 465, 865–877.
Elitzur, M. (1976), ‘Inversion of the OH 1720-MHz line’, ApJ 203, 124–131.
Elitzur, M. (1992), Astronomical Masers, Dordrecht Netherlands.
Elitzur, M. (1998), ‘Polarization of astronomical maser radiation IV. Circular polarization pro-
files’, ApJ 504, 390–395.
Ellingen, S. P., Breen, S., Voronkov, M., Caswell, J., Chen, X. and Titmarsh, A. (2011), An
evolutionary timeline for high mass star formation, in ‘Science with Parkes @ 50 Years
Young, 31 Oct. – 4 Nov., 2011’.
Ellingsen, S. (2006), ‘Methanol masers: reliable tracers of the early stages of high-mass star
formation’, ApJ 638, 241–261.
Farmer, H. (2014), ‘Disk-outflow models as applied to high mass star forming regions through
methanol and water maser observations’, IJAA 4, 571–597.
Feigelson, E. D. and Montmerle, T. (1999), ‘High-energy processes in Young Stellar Objects’,
ARA&A 37, 363–408.
Fish, V. L. and Reid, M. J. (2006), ‘Full- polarization observation of OH masers in mas-
sive star forming regions. II. Maser properties and the interpretation of polarization’, ApJS
164, 99–123.
Fish, V. L., Reid, M. J., Argon, A. L. and Zheng, X.-W. (2005), ‘Full- polarization observation
of OH masers in massive star forming region. I. Data’, ApJS 160, 220–271.
Forster, J. and Caswell, J. (2000), ‘Radio continuum emission at OH and H2O maser sites’,
ApJ 530, 371–386.
Franco, J., Kurtz, S., Hofner, P., Testi, L., Garcia-Segura, G. and Martos, M. (2000), ‘The
density structure of highly compact HII regions’, ApJ 542, L143–L146.
Garay, G., Rodriguez, L. F., Moran, J. M. and Churchwell, E. (1993), ‘VLA observation of
strong IRAS point sources associated with compact HII regions’, ApJ 418, 368.
Garcia-Barreto, J. A., Burke, B. F., Reid, M. J., Moran, J. M., Haschick, A. D. and Schilizzi,
R. T. (1988), ApJ 325, 954–966.
Genzel, R., Reid, M., Moran, J. and Downes, D. (1981), ‘Proper motions and distances of H2O
maser sources. I—The outflow in Orion-KL’, ApJ 244, 884.
Goedhart, S., Gaylard, M. J. and van der Walt, D. J. (2003), ‘Periodic flares in the methanol
maser source G9.62+0.20E’, MNRAS 339, L33–L36.
Goedhart, S., Gaylard, M. J. and van der Walt, D. J. (2004), ‘Long term monitoring of 6.7 GHz
methanol masers’, MNRAS 355, 553–584.
Goedhart, S., Maswanganye, J. P., Gaylard, M. J. and van der Walt, D. J. (2014), ‘Periodicity
in class II methanol masers in high-mass star-forming regions’, MNRAS 437, 1808–1820.
Goedhart, S., van Rooyen, R., van der Walt, D. J., Maswangangye, J., Sanna, A., Macleod,
G. and van den Heever, S. (2019), ‘Periodic variability of the mainline hydroxyl masers
G9.62+0.20E’, MNRAS 485, 4676–4685.
185
 University of Ghana http://ugspace.ug.edu.gh
Green, J. A., Caswell, J. L., Fuller, G. A., Avison, A., Breen, S. L., Ellingsen, S. P., Pestalozzi,
M., Quinn, L., Thompson, M. A. and Voronkov, M. A. (2010), ‘The 6-GHz methanol multi-
beam maser catalogue – II. Galactic longitudes 6 ◦ to 20 ◦’, MNRAS 409(3), 913–935.
Green, J. A., Caswell, J. L., Voronkov, M. and Mcclure-Griffths, N. M. (2012), ‘Variability
monitoring of the hydroxyl maser emission in G12.889+0.489’, MNRAS 425, 1504–1510.
Greene, T. (2001), ‘Protostars: "Stellar embryology" takes a step forward with the
first detailed look at the youngest Sun-like stars’, American Scientist 89, 316–325.
Guzman, A., Garay, G., Rodrıguez, L., Contreras, Y., Dougados, C. and Cabrit, S. (2016), ‘A
protostellar jet emanating from a hypercompact HII region’, ApJ 826, 208.
Hacar, A., Tafalla, M., Kauffmann, J. and Kovacs, A. (2013), ‘Cores, filaments, and bundles:
hierarchical core formation in the L1495/B213 Taurus region.’, A&AS 554, A55.
Hatchell, J., A., T. M., Millar, T. J. and Macdonald, G. (1998), ‘A survey of molecular line
emission towards ultracompact HII regions’, A&AS 133(1), 29–49.
Heywood, I., Klöeckner, H., Beswick, R. Garrington, S. T., Hatchell, J., Hoare, M. G., Jarvis,
M. J., Jones, I., Muxlow, T. W. B. and S., R. (2011), ‘Expanding e-MERLIN with the Goon-
hilly Earth Station’, arXiv:astro-ph.IM/1103.1214 .
Hill, T., Minier, V., Thompson, M. A., Walsh, A. J., Burton, M. G., Huny- Cunningham, M. and
Garay, G. (2005), ‘Millimetre continuum observations of southern massive star formation
regions – I. SIMBA observations of cold cores’, A&A 429, 946–960.
Hirota, T., Bushimata, T. and Choi, Y. K. (2007), ‘Distance to Orion KL measured with VERA’,
PASJ 59, 897.
Hoare, M. G. (2002), ‘Hot Star Workshop 111:The earliest phases of massive star birth’, ASP
Conf. Ser. 267, 137.
Hoare, M. G. and Franco, J. (2007), ‘Massive star formation’, arXiv:astro-ph 0711.4912v1 .
Hoare, M. G., Lumsden, S. L. and Oudmaijer, R. D. (2005), Hypercompact HII regions,IAU
Symposium proceedings of the International Astronomical Union massive star birth: A
crossroads of astrophysics, Vol. 227, Cambridge University Press, Cambridge.
Hoare, M. G.and Kurtz, S. E., Lizano, S., Keto, E. and Hofner, P. (2007), Ultra-compact HII
regions and the early lives of massive stars, in B. K. K. and J. David, eds, ‘Protostars and
Planets V’, Vol. V, University of Arizona Press, Tucson, Arizona, p. 181–196.
Hofner, P., Kurtz, S., Churchwell, E. and Walmsley, C. M.and Cesaroni, R. (1996), ‘Massive
star formation in the hot dense cloud of G9.62+0.19’, ApJ 460, 359.
Hofner, P., Wiesemeyer, H. and Henning, T. (2001), ‘A high velocity molecular outflow from
the G9.62+0.19 star forming region’, ApJ 549, 425.
Högbom, J. A. (1974), ‘Aperture synthesis with a non-regular distribution of interferometer
baselines’, A&AS 15, 417.
Hosokawa, T. and Omukai, K. (2009), ‘Evolution of massive protostars with high accretion
rates’, ApJ 691, 823–846.
Hosokawa, T., Yorke, H. W. and Omukai, K. (2010), ‘Evolution of massive protostars via disk
accretion’, ApJ 721, 478.
186
 University of Ghana http://ugspace.ug.edu.gh
Hoyle, F. (1954), ‘On nuclear reactions occuring in very hot stars.I. The synthesis of elements
from carbon to nickel’, ApJS 1, 121.
Inayoshi, K., Sugiyama, K., Hosokawa, T., Motogi, K. and Tanaka, K. E. I. (2013), ‘Direct
diagnostics of forming massive stars: Stellar pulsation and periodic variability of maser
sources’, ApJL 769(L20), 5.
Isella, A. (2006), Interferometric observations of pre-main sequence disks, Phd, Universit‘a
degli Studi di Milan, Milan, Italy.
Jackson, J. D. (1999), Classical Electrodynamics, 3rd edn, New York.
Johnston, K. G., Robitaille, T. P., Beuther, H., Linz, H., Boley, P.Kuiper, R., Keto, E., Hoare,
M. G. and Boeke, R. v. (2015), ‘A keplerian-like disk around the forming O-type star AFGL
4176’, ApJL 813, L19.
Keto, E. (2007), ‘The formation of massive stars: Accretion, disks, and the developement of
hypercompact HII regions’, ApJ 666, 976–981.
Kolligan, A. and Kuiper, R. (2017), ‘From accretion to outflows of massive protostars.’, Mem.
Soceita Astronomica Italiana 88, 793.
Krumholz, M. (2014), ‘The big problems in star formation: The star formation rate, stellar
clustering, and the initial mass function’, Phys. Rep. 539, 49–134.
Krumholz, M. (2015a), The formation of very massive stars, in ‘Very Massive Stars in the
Local Universe’, Vol. 412 of Ap&SSL, p. 43.
Krumholz, M. (2015b), ‘Notes on star formation’, arXiv e-prints .
Krumholz, M. R., Cunningham, A. J., Klein, R. I. and Mckee, C. F. (2010), ‘Radiation feed-
back, fragmentation, and the environmental dependence of the initial mass function’, ApJ
713, 1120–1133.
Kuiper, R., Klahr, H., Beuther, H., Henning, T. and Yorke, H. (2013), Radiation pressure in
massive star formation, in B. K. K. and J. David, eds, ‘Protostars and Planets VI’, University
of Arizona Press, Tucson, Arizona.
Kuiper, R., Turner, N. and Yorke, H. (2016), ‘Protostellar Outflows and Radiative Feedback
from Massive Stars. II. Feedback, Star-formation Efficiency, and Outflow Broadening’, ApJ
832, 40.
Kurtz, S. (2005), Hypercompact HII regions,massive star birth: A crossroads of astrophysics,
iau symposium proceedings of the international astronomical union, Vol. 227, Cambridge
University Press, Cambridge, pp. 111–119.
Kurtz, S., Cesaroni, R., Churchwell, E., Hofner, P. and Walmsley, C. M. (2000), ‘Hot molec-
ular cores and the earliest phases of high-mass star formation’, Protostars and Planets IV
pp. 299–326.
Kurtz, S., Churchwell, E. and Wood, D. O. S. (1994), ‘Ultracompact HII regions: New high-
resolution radio images’, ApJS 91, 659–712.
Kurtz, S. and Franco, J. (2002), ‘Ultracompact HII regions’, Rev. Mex. AA (Serie de Conferen-
cias) 12, 16–21.
Lada, C. (1985), Star formation : From OB associations to protostars, in ‘IAU Symposium’,
Vol. 115, Dordrecht, D. Reidel Publishing Co, pp. 1–18.
187
 University of Ghana http://ugspace.ug.edu.gh
Linz, H., Stecklum, B., Henning, T. and Hofner, P. and. Brandl, B. (2005), ‘The G9.62+0.19 F
hot molecular core : The infrared view on very young massive stars’, A&A 429(3), 903–921.
Liu, T., Lacy, J., Shing Li, P., Wang, K., Qin, S., Zhang, Q., Kim, K. andGaray, G., Wu,
Y., Mardones, D., Zhu, Q. andTatematsu, K. H. T. and Ren, Z. andLu, S. a. H. a. Y. L. D.
(2017), ‘ALMA reveals sequential high mass star formation in the G9.62+0.19 complex’,
ApJ 849, 25–46.
Liu, T., Wu, Y., Liu, S.-Y., Qin, S.-L., Su, Y.-N., Chen, H.-R. and Ren, Z. (2011), ‘Infall and
outflows motions in the high- mass star-forming complex G9.62+0.19’, ApJ 730(102), 17.
Loinard, L., Torres, R., Mioduszewski, A. and Rodriguez, L. (2008), ‘A preliminary VLBA
distance to the core of Ophiuchus, with an accuracy of 4%’, ApJ 675, L29–L32.
Luhman, K. L., Rieke, G. H., Lada, C. J. and Lada, E. (1998), ‘Low mass star formation and
the initial mass function in IC 348’, ApJ 508, 347–369.
Macleod, G. and Gaylard, M. (1996), ‘Variable hydroxyl and methanol masers in G351.78-
0.54’, MNRAS 280, 868–876.
Marr, J. M., Snell, R. L. and Kurtz, S. E. (2016), Fundamentals of Radioastronomy : Observa-
tional methods, Series in Astronomy and Astrophysics.
Marsh, K. A. (1975), ‘The radio spectra of mass-outflow stars’, ApJ 201, 190–193.
Maud, L. T., Cesaroni, R., Kumar, M. S., Van Der Tak, F. F., Allen, V., Hoare, M. G., Klaassen,
P. D., Harsono, D.and Hogerheijde, M. R., Sánchez-Monge, A., Schilke, P., Ahmadi, A.,
Beltrán, M. T., Beuther, H., T., C., Etoka, S., Fuller, G.and Galván-Madrid, R., Goddi, C.,
Henning, T. and Johnston, K. G. (2018), ‘Chasing discs around O-type (proto)stars: ALMA
evidence for an SiO disc and disc wind from G17.64+0.16’, A&A 620, A31.
McGuire, C., Fuller, G., Peretto, N., Zhang, Q., Traficante, A., Avison, A. and Jimenez-Serra,
I. (2016), ‘The structure and early evolution of massive star forming regions. Substructure in
the infrared dark cloud SDC13’, A&A 594, A118.
McKee, C. F. and Tan, J. C. (2003), ‘The formation of massive stars from turbulent cores’, ApJ
585, 850–871.
Mcmullin, J. (2009), CASA synthesis and single dish Reduction cookbook, beta release edition
edn.
Menten, K. M. (1991), ‘The discovery of a new, very strong, and widespread interstellar
methanol maser line’, ApJL 380, L75.
Meyers, A. T. and Mckee, C. F. (2013), ‘The fragmentation of magnetized, massive star-
forming cores with radiative feedback’, ApJ 766, 97–115.
Mezger, P. and Henderson, A. (1967), ‘Galactic HII regions I Observations of their continuum
radiation at the frequency 5 GHz’, ApJ 147, 471–489.
Minier, V., Burton, M., Hill, T., Pestalozzi, M., Purcell, C. R., Garay, G. and Walsh, J. (2005),
‘Star forming protoclusters associated with methanol masers’, A&A 429, 945.
Minier, V., Conway, J. E. and Booth, R. (2001), ‘VLBI observations of 6.7 and 12.2 GHz
methanol masers towards high mass star forming regions: Tracing massive protostars.’, A&A
369, 278–290.
188
 University of Ghana http://ugspace.ug.edu.gh
Minier, V., Ellingsen, S. P., Norris, R. P. and Booth, R. S. (2003), ‘The protostellar mass limit
for 6.7 GHz methanol masers. I. A low-mass YSO survey’, A&A 403, 1095–1100.
Morton, D. C. (1967), ‘The Far-ultraviolet spectra of six stars in Orion’, AJ 147, 1017.
Moscadelli, L. and Reid, M. J.and Menten, K. M. (2009), ‘Trigonometric parallaxes of massive
star-forming regions. II. Cep A and NGC 7538’, ApJ 693, 406.
Offringa, A., De Bruyn, A., Biehl, M., Zaroubi, S., Bernardi, G. and Pandey, V. (2010), ‘Post-
correlation radio frequency interference classification methods’, MNRAS 405, 155–167.
Oozer, N., Bietenholz, M. and Goedhart, S. (2010), An introduction to CASA: A KAT-7 Data
reduction guide, SKA, SA.
Ott, M., Witzel, A., Quirrenbach, A., Krichbaum, T., Standke, K., Schalinski, C. and Hummel,
C. (1994), ‘An updated list of radio flux density calibrators’, A&A 284, 331–339.
Parfenov, S. Y. and Sobolev, A. M. (2014), ‘On the class II methanol maser periodic variability
due to the rotating spiral shocks in the gaps of discs around young binary stars’, MNRAS
444, 620.
Peck, L. and Fenech, D. M. (2013), ‘SERPent: Automated reduction and RFI-mitigation soft-
ware for e-merlin’, Astron and Comput 2, 54–66.
Peretto, N. and Fuller, G. A. (2009), ‘The initial conditions of stellar protocluster formation. I.
A catalogue of spitzer dark clouds’, A&A 505, 405–415.
Perley, R. A. and Butler, B. J. (2013), ‘An accurate flux density from 1 to 50 GHz’, ApJS
204, 19–39.
Pestalozzi, R., Elitzur, M., Conway, J. E. and Booth, R. S. (2004), ‘A circumstellar disk in a
high mass star forming region’, ApJ 603, L113–L116.
Purcell, C., Balasubramanyam, R., Burton, M., Walsh, A., Minier, V., Hunt-Cunningham, M.,
Kedziora-Chudczer, L., Longmore, S., Hill, T., Bains, I., Barnes, P., Busfield, A., Calisse,
P., Crighton, N., Curran, S., Davis, T., Dempsey, J., Derragopian, G., Fulton, B., Hidas,
M., Hoare, M. and Lee, J. K. Lumsden, S. (2006), ‘A CH3CN and HCO+ survey towards
southern methanol masers associated with star formation’, MNRS 367(2), 553–576.
Purser, S., Lumsden, S., Hoare, M., Urquhart, J., S., Cunningham, N., Purcell, C.R.and Brooks,
K. G. G. a. A. and Voronkov, M. (2016), ‘A search for ionized jets towards massive young
stellar objects’, MNRAS 460, 1039–1053.
Reid, M. J. and Moran, J. M. (1981), ‘Masers’, ARA&A 19, 231–276.
Sakai, T., Sakai, N., Hirota, T. and Yamamoto, S. (2010), ‘A survey of molecular lines towards
massive clumps in early evolutionary stages of high-mass star formation’, ApJ 714, 1658–
1671.
Sanchez-Monge, A., Beltran, M., Cesaroni, R.and Etoka, S., Galli, D., Kumar,
M.S.N.and Moscadelli, L., Stanke, T.and van der Tak, F., Vig, S., Walmsley, C., Wang,
K., Zinnecker, H., Elia, D., Molinari, S. and Schisano, E. (2014), ‘A necklace of dense cores
in the high-mass star forming region G35.20-0.74 N: ALMA observations’, A&A 569, A11.
Sanna, A. (2019), ‘Size of the G9.62+0.20E HII region at Q band’, Private email communica-
tion.
189
 University of Ghana http://ugspace.ug.edu.gh
Sanna, A., Menten, K. M., Carreasco-Gonzalez, C., Reid, M. J., Ellingsen, S. P., Brunthaler,
A. and Moscadelli, L. (2015), ‘The environment of the strongest galactic methanol masers’,
ApJL 804, L2.
Sanna, A., Reid, M. J., Moscadelli, L., Dame, T. M., Menten, K. M., Brunthaler, A., Zheng,
X. W. and Xu, Y. (2009), ‘Trigonometric parallaxes of massive star forming regions. VII.
G9.62+0.20 and the expanding 3 kpc arm’, ApJ 706, 464–470.
Shu, F. H., Adams, F. C. and Lizano, S. (1987), ‘Star formation in molecular clouds: observa-
tion and theory’, ARA&A 25, 23–81.
Slysh, V. I., Kalenskii, S. V. and Val’tts, I. E. (2002), ‘Methanol radio emission at millimeter
wavelengths: New masers at 1.3 and 2.8 millimeters’, Astron. Rep. 46(1), 49–56.
Smith, R. J., Glover, S. C. O. and Klessen, R. S. (2014), ‘On the nature of star-forming filaments
- I. filament morphologies’, MNRAS 445, 2900–2917.
Sobolev, A. M., Cragg, D., Ellingsen, S., Gaylard, M., Goedhart, S., Henkel, C. and Voronkov,
M. (2007), ‘How do methanol masers manage to appear in the youngest star vicinities and
isolated molecular clumps?’, Proc. IAU Symp. 3(S242), 81–88.
Sridharan, T., Williams, S. J. and Fuller, G. A. (2005), ‘The direct detection of a (proto) bi-
nary/disk system in IRAS 20126+4104’, ApJL 631, L73.
Stahler, S. W. and Palla, F. (2004), The formation of Massive Stars, 1st edn, Wiley-VCH.
Strömgren, B. (1939), ‘The physical state of interstellar hydrogen’, ApJ 89, 526–547.
Sullivan, W. T. and Kerstholts, J. (1976), ‘Time variation in the 18cm OH emission profiles
over the period 1965-1972’, A&A 51, 427–450.
Svoboda, B. E., Shirley, Y. L., Alessio, T., Cara, B., Fuller, G. A., Qizhou, Zhang and Henrik,
B., Peretto, N., Brogan, C. and Hunter, T. (2019), ‘ALMA Observations of fragmentation,
substructure, and protostars in high-mass starless slump candidates’, ApJ 886(1), 36.
Swift, J. J. and Welch, J. W. (2008), ‘A case study of low mass star formation’, ApJSS 174, 202–
222.
Szymczak, M., Blaszkiewicz, L., Etoka, S. and M., L. A. (2001), ‘Variations of polarisation
properties of OH maser emission from three semiregular variables’, A&A 379(3), 884–892.
Tan, J. C., Beltran, M. T., Caselli, P., Fontani, F., Fuente, A., Krumholz, M. R., Mckee, C. F.
and Solte, A. (2014), Massive star formation, in B. K. K. and J. David, eds, ‘Protostars and
Planets’, Vol. VI, University of Arizona Press, Tucson, Arizona.
Tennyson, J. (2003), Molecules in space, Vol. 3 of Handbook of Molecular Physics and Quan-
tum chemistry, John Wiley & Sons, chapter 14.
Testi, L., Felli, M. and Persi, P. andRoth, M. (1998), ‘HII region and hot dust emission around
young massive stars in G9.62+0.19’, ApJ 508, L91–L94.
Testi, L., Hofner, P., Kurtz, S. and Rupen, M. (2000), ‘Detection of the thermal radio continuum
emission from the G9. 62+ 0.19-F Hot Core’, A&A 359, l5.
Thompson, A. R., Moran, J. M. and Swenson, G. W. (2001), Interferometry and synthesis in
radio astronomy, 2nd edn, Wiley, New York.
Tielens, A. G. G. M. (2013), ‘The molecular universe’, Rev. Mod. Phys. 85, 1021.
190
 University of Ghana http://ugspace.ug.edu.gh
Traficante, A., Fuller, G. A., Peretto, N., Pineda, J. E. and Molinari, S. (2015), ‘The initial con-
ditions of stellar protocluster formation - II. A catalogue of starless and protostellar clumps
embedded in IRDCs in the galactic longitude range 15deg≤l≤55deg′,MNRAS451, 3089−
−3106.
Traficante, A., Fuller, G., Duarte Cabral, A., Elia, D.and Heyer, M. H., Molinari, S., Peretto,
N. and Schisano, E. (2020), ‘Multi-scale dynamics in star-forming regions: the interplay
between gravity and turbulence’, MNRAS 491, 4310–4324.
Trung, D.-V. (2009), ‘On the theory of astronomical maser II. Polarization of maser radiation’,
MNRAS 399, 1495–1505.
Urquhart, J., Figura, C., Moore, T., Csengeri, T., Lumsden, S., Pillai, T., Thompson, M., Eden,
D. and Morgan, L. (2015), ‘The rms survey: ammonia mapping of the environment of mas-
sive young stellar objects’, MNRAS 452, 4029.
Urquhart, J. S., Busfield, A. L., Hoare, M. G., Lumsden, S. L., Oudmaijer, R. D., Moore, T.
J. T., Gibbs, A. G., Purcell, C. R., Burton, M. G., Marrechal, W. L.and Jiang, Z. and Wang,
M. (2008), ‘The RMS survey: CO Observation of the candidate massive YSO in the northern
galactic plane’, A&A 487, 253–264.
Urquhart, J. S., Thompson, M. A., Moore, T. J. T., Purcell, C. R., Hoare, M. G., Schuller, F.,
Wyrowski, F., Csengeri, T., Menten, K. M., Lumsden, S. L., Kurtz, S., Walmsley, C. M.,
Bronfam, L., Morgan, L. K., Eden, D. J. and Russeil, D. (2013), ‘ATLASGAL – Properties
of compact HII regions and their natal clumps’, MNRAS 435, 400–428.
van den Heever, S. P., van der Walt, D., Pittard, J. and Hoare, M. (2019), ‘Periodic methanol
masers from a colliding wind binary perspective’, MNRAS 485, 2759–2771.
van der Walt, D. (2011a), ‘On the methanol masers in G9.62+0.20E and G188.95 + 0.89’, AJ
141, 152.
van der Walt, D. (2011b), ‘On the periodic class II methanol masers in the high mass star form-
ing region G9.62+0.20E’, Bulletin del a Societe Royale des Science de Liège 80, 260–264.
van der Walt, D., , Goedhart, S. and Gaylard, M. J. (2009), ‘Periodic class II methanol masers
in G9.62+0.20E’, MNRAS 398, 961.
van Dishoeck, E. F. and Blake, G. A. (1998), ‘Chemical evolution of star forming regions’,
ARA&A 36, 317–368.
Viti, S. and Williams, D. A. (1999), ‘Time-dependent evaporation of icy mantles in hot cores’,
MNRAS 305, 755–762.
Viti, S., Williams, D. A. and O’Neill, P. (2000), ‘Hydrocarbons in diffuse and translucent
clouds’, A&A 354, 1062–1070.
Walsh, A. J., Burton, M. G., Hyland, A. R. and Robinson, G. (1999), ‘Studies of ultracompact
HII regions-III. near infrared survey of selected regions’, MNRAS 309, 905–922.
Walsh, A. J., Burton, M., Hyland, A. R. and Robinson, G. (1998), ‘Studies of ultracom-
pact HII region II. high resolution radio continuum and methanol maser survey’, MNRAS
301, 640–698.
Walsh, A. J., Hyland, A. R., Robinson, G. and Burton, M. G. (1997), ‘Studies of ultracompact
HII regions - I. Methanol maser survey of lRAS-selected sources’, MNRAS 291, 261–278.
191
 University of Ghana http://ugspace.ug.edu.gh
Wang, J. W., Lai, S. P., Chakali, E., Kate, P., Francesco, J. D., Doug, J., Koch, P. M., Tie, L.,
Motohide, T., Furuya, R. S., Takashi, O., Ward-Thompson, D., Soam, A., Kim, K., Lee, C.,
Lee, C., Mairs, S., Arzoumanian, D., Kim, G., Hoang, T., Hwang, J., Liu, S., Berry, D.,
Bastien, P., Hasegawa, T., Kwon, W., Qiu, K. and André, P. (2019), ‘JCMT BISTRO Survey:
Magnetic fields within the hub-filament structure in IC 5146’, ApJ 876, 42.
Wang, K., Zhang, Q., Testi, L., van der Tak, F., Wu, Y. and Zhang, H. (2014), ‘Hierarchial
fragmentation and differential star formation in the Galactic ’Snake’: infrared dark cloud
G11. 11-0.12’, MNRAS 439, 3275–3293.
Willacy, K. and Williams, D. (1993), ‘Desorption processes in molecular clouds:quasi steady
state chemistry’, MNRAS 260, 635–642.
Wilson, T. L., Rohlfs, K. and Hüttemeister, S. (2009), Tools of radio astronomy, 2nd edn,
Springer, Berlin.
Wood, D. O. S. and Churchwell, E. (1989), ‘The morphologies and physical properties of
ultracompact Hii regions’, ApJS 69, 831–895.
Yamamoto, S. (2017), Introduction to Astrochemistry: Chemical evolution from interstellar
clouds to star formation, 1st edn, Springer Japan.
Yorke, H. W. and Sonnhalter, C. (2002), ‘On the formation of massive stars’, ApJ 569, 846–862.
Zhang, Q., Wang, K. and Lu, X. (2015), ‘Fragmentation of molecular clumps and formation of
a protocluster’, ApJ 804, 141.
Zinnecker, H. (1982), Symposium on the Orion Nebula to Honour Henry Draper. NY Acad. Sci.
p. 226.
Zinnecker, H. and Yorke, H. W. (2007), ‘Toward understanding massive star formation’,
ARA&A 45, 481.
192