© 2020 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society
ORIGINAL UNEDITED MANUSCRIPT
Type-I X-ray Bursts from MAXI J1807+132 1
Discovery of Thermonuclear Type-I X-ray Bursts from the
X-ray binary MAXI J1807+132
A. C. Albayati,
1?
D. Altamirano,
1
G. K. Jaisawal,
2
P. Bult,
3,4
S. Rapisarda,
5
G. C. Mancuso,
6,7
T. G
¨
uver,
8,9
Z. Arzoumanian,
4
D. Chakrabarty,
10
J. Chenevez,
2
J. M. C. Court,
11
K. C. Gendreau,
4
S. Guillot,
12,13
L. Keek,
14
C. Malacaria,
15,16
and T. E. Strohmayer
4,17
1
School of Physics and Astronomy, University of Southampton, Southampton, SO17 1BJ, UK
2
National Space Institute, Technical University of Denmark, Elektrovej 327-328, DK-2800 Lyngby, Denmark
3
Department of Astronomy, University of Maryland, College Park, MD 20742, USA
4
Astrophysics Science Division, NASA’s Goddard Space Flight Center, Greenbelt, MD 20771, USA
5
Shanghai Astronomical Observatory, Chinese Academy of Sciences, 80 Nandan Road, Shanghai 200030, China
6
Instituto Argentino de Radioastronom´ıa (CCT-La Plata, CONICET; CICPBA), C.C. No. 5, 1894 Villa Elisa, Argentina
7
Facultad de Ciencias Astron´omicas y Geof´ısicas, Universidad Nacional de La Plata, Paseo del Bosque s/n, 1900 La Plata, Argentina
8
Istanbul University, Science Faculty, Department of Astronomy and Space Sciences, Beyazıt, 34119, Istanbul, Turkey
9
Istanbul University Observatory Research and Application Center, Istanbul University 34119, Istanbul Turkey
10
MIT Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology, Cambridge, MA 02139, USA
11
Department of Physics and Astronomy, Texas Tech University, PO Box 41051, Lubbock, TX 79409, USA
12
CNRS, IRAP, 9 avenue du Colonel Roche, BP 44346, F-31028 Toulouse Cedex 4, France
13
Universit´e de Toulouse, CNES, UPS-OMP, F-31028 Toulouse, France
14
cosine B.V., Oosteinde 36, 2361 HE Warmond, The Netherlands
15
NASA Marshall Space Flight Center, NSSTC, 320 Sparkman Drive, Huntsville, AL 35805, USA
16
Universities Space Research Association, Science and Technology Institute, 320 Sparkman Drive, Huntsville, AL 35805, USA
17
Joint Space-Science Institute, NASA’s Goddard Space Flight Center, Greenbelt, MD 20771, USA
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
MAXI J1807+132 is a low-mass X-ray binary (LMXB) first detected in outburst in 2017. Ob-
servations during the 2017 outburst did not allow for an unambiguous identification of the nature of
the compact object. MAXI J1807+132 was detected in outburst again in 2019 and was monitored
regularly with NICER. In this paper we report on five days of observations during which we de-
tected three thermonuclear (Type-I) X-ray bursts, identifying the system as a neutron star LMXB.
Time-resolved spectroscopy of the three Type-I bursts revealed typical characteristics expected for
these phenomena. All three Type-I bursts show slow rises and long decays, indicative of mixed H/He
fuel. We find no strong evidence that any of the Type-I bursts reached the Eddington Luminosity;
however, under the assumption that the brightest X-ray burst underwent photospheric radius ex-
pansion, we estimate a < 12.4 kpc upper limit for the distance. We searched for burst oscillations
during the Type-I bursts from MAXI J1807+132 and found none (< 10% amplitude upper limit
at 95% confidence level). Finally, we found that the brightest Type-I burst shows a ∼ 1.6 sec pause
during the rise. This pause is similar to one recently found with NICER in a bright Type-I burst
from the accreting millisecond X-ray pulsar SAX J1808.4–3658. The fact that Type-I bursts from
both sources can show this type of pause suggests that the origin of the pauses is independent of
the composition of the burning fuel, the peak luminosity of the Type-I bursts, or whether the NS is
an X-ray pulsar.
Key words: stars: neutron – stars: individual (MAXI J1807+132) – X-rays: binaries
– X-rays: bursts
?
E-mail: a.c.albayati@soton.ac.uk
1 INTRODUCTION
Low-mass X-ray binaries (LMXBs) consist of either a neu-
tron star (NS) or a black hole (BH) primary accreting ma-
Downloaded from https://academic.oup.com/mnras/advance-article/doi/10.1093/mnras/staa3657/6000262 by DTU Library user on 26 November 2020
ORIGINAL UNEDITED MANUSCRIPT
2 A. C. Albayati et al.
terial from a low mass (. 1 M
) companion star through
Roche lobe overflow. Gas flowing from the companion star
forms an accretion disc around the primary. As the gas in the
disc spirals closer to the compact object, gravitational poten-
tial energy is released in the form of X-rays. A LMXB spends
most of its life in a quiescent (very low or non-accreting)
state, but when in outburst, it can reach X-ray luminosities
of L
X
' 10
34−38
erg s
−1
(see, e.g., Tauris & van den Heuvel
2006, for a review).
It is challenging to differentiate between LMXBs host-
ing a BH or NS. Dynamical measurements of a compact ob-
ject’s mass can be used to help identify its nature, however,
the sensitivity of the observations needed to make these esti-
mations is difficult to acquire (e.g., Casares & Jonker 2014).
The detection of coherent pulsations confirms a NS identi-
fication, as these are associated with the spin frequency of
the NS (e.g., Patruno & Watts 2012). Similarly, thermonu-
clear burning requires a solid surface for the fuel to settle
on. A BH has no solid surface, but an event horizon instead,
hence the detection of thermonuclear burning also secures
a NS identification. There are additional observables that
can be used to identify the nature of the compact object,
although the identification is based on empirical evidence
that certain phenomenology has only been observed in ei-
ther a BH or a NS so far. For example, quasi-periodic os-
cillations (QPOs) with frequencies higher than 500 Hz have
only been seen in X-ray light curves from observations of
NS systems (see, e.g., van der Klis 2006, for a review). How-
ever, the observation and characterisation of QPOs, and
the power-spectral broadband noise, with frequencies be-
low a few hundred Hertz are not always conclusive (e.g.,
Klein-Wolt & van der Klis 2008). Multi-wavelength obser-
vations have been also used to distinguish between BHs and
NSs. For example, NSs can be ∼ 30 times fainter in the ra-
dio band when compared with BHs observed at similar X-ray
luminosities. However, recent works have shown that there
is a population of radio-faint BHs that have similar radio
luminosities to NSs (Tetarenko et al. 2018).
Thermonuclear burning in a NS atmosphere can
manifest as stable, unstable, or marginally stable (e.g.,
Galloway & Keek 2017). Here, we concentrate on unstable
thermonuclear (Type-I) X-ray bursts (hereafter referred to
as “X-ray bursts”). X-ray bursts appear as a sudden in-
crease in X-ray emission over timescales of seconds. They
occur when pure or mixed material - Hydrogen, Helium, and
sometimes Carbon - accreted onto the NS surface reaches
a critical density and temperature which allows for run-
away thermonuclear burning (see, e.g., Lewin et al. 1993;
Strohmayer & Bildsten 2003, for reviews). During this pro-
cess, the X-ray flux increases rapidly (. 1 − 10 s) and is
generally followed by an exponential-like decay over tens of
seconds as the NS atmosphere cools.
When a Type-I X-ray burst reaches the Eddington limit,
L
Edd
, it results in photospheric radius expansion (PRE, see,
e.g., Tawara et al. 1984; Lewin et al. 1984), where the out-
ward radiation pressure exceeds the inward gravitational
force. Since black body luminosity scales as L ∝ R
2
T
4
, when
the X-ray burst reaches the Eddington limit an increase in
radius and decrease in temperature can be seen in time-
resolved spectral analysis of PRE X-ray bursts. An X-ray
burst’s luminosity remains roughly constant at L
Edd
dur-
Table 1. Details of the NICER observations analysed in this
paper. Quoted exposure times are as after processing.
# ObsID Start Time Date Exposure X-ray
(MJD) (DD-MM-YY) (sec) Burst
1 2200840122 58783.04946 27-10-2019 15838 -
2 2200840123 58784.01681 28-10-2019 12557 Yes
3 2200840124 58785.04902 29-10-2019 5839 Yes
4 2200840125 58786.27471 30-10-2019 4480 Yes
5 2200840126 58787.04898 31-10-2019 4850 -
ing PRE, thus PRE X-ray bursts can be used as empirical
standard candles (van Paradijs 1978; Kuulkers et al. 2003).
MAXI J1807+132 (hereafter MAXI J1807) was first
discovered by the nova-alert system of the Monitor of
All-sky X-ray Image Gas Slit Camera (MAXI /GSC;
Matsuoka et al. 2009; Mihara et al. 2011) during its 2017
outburst (Negoro et al. 2017). This detection was followed
up with The Neil Gehrels Swift Observatory (Swift; Gehrels
2004) X-Ray Telescope (XRT; Burrows et al. 2003) obser-
vations, spectral analysis of which suggested a LMXB with
a NS primary (Shidatsu et al. 2017). Further X-ray stud-
ies with XMM-Newton’s European Photon Imaging Cam-
era (EPIC; Str
¨
uder et al. 2001) and ground-based opti-
cal telescopes supported the NS identification, although
the possibility of a BH primary could not be ruled out
(Jim´enez-Ibarra et al. 2019).
After roughly 2 years in quiescence, a new out-
burst of MAXI J1807 was detected on 10 Septem-
ber 2019 (MJD 58736) by the MAXI /GSC nova-
alert system (Shidatsu et al. 2019). Subsequently, the
Neutron Star Interior Composition Explorer (NICER;
Gendreau & Arzoumanian 2017) monitored the source from
16 September (MJD 58742), and observed the system
on a regular basis. The outburst of MAXI J1807 was
characterised by flaring events lasting days to weeks
(Rapisarda et al. 2019). Two X-ray bursts were detected by
NICER on 28 and 29 October, and preliminary analysis of
these two events reported by Arzoumanian et al. (2019) con-
firmed MAXI J1807 as a neutron star LMXB. In this pa-
per we report on the detection of a third X-ray burst with
NICER, and present a detailed analysis of all three Type-I
bursts.
2 OBSERVATIONS AND DATA ANALYSIS
NICER is an X-ray telescope on board the International
Space Station. It was launched in 2017 with instrumentation
specifically designed for the study of NSs. NICER’s X-ray
Timing Instrument (XTI) operates in the 0.2−12 keV energy
band, providing high timing and spectral resolution of 100 ns
and 6 < E/∆E < 80 from 0.5 keV to 8 keV respectively.
With 52 active detectors, NICER provides an effective area
of 1900 cm
2
at 1.5 keV.
NICER observed MAXI J1807 between 16 September
and 26 November 2019, generating a total of 47 observa-
tion IDs (ObsIDs). We searched all available data for X-ray
bursts; here we report on the five observations around the
time of the detection of three X-ray bursts (see Table 1).
The observations correspond to ObsIDs 2200840122–26, per-
Downloaded from https://academic.oup.com/mnras/advance-article/doi/10.1093/mnras/staa3657/6000262 by DTU Library user on 26 November 2020
ORIGINAL UNEDITED MANUSCRIPT
Type-I X-ray Bursts from MAXI J1807+132 3
0 2.8 5.6 8.4
11 14 17
20
22
25 28
07:45.0
18:08:00.0
15.0
30.0
08:00.0 12:00.0 16:00.0 13:20:00.0
Figure 1. A 3–78 keV NuSTAR image of the NICER field-of-
view, 26 September 2019 (ObsID 90501342002, exposure time of
22 ks). The NICER field-of-view for our observations is denoted
by a green circle of 3 arcmin radius.
formed between 27 October (MJD 58783) and 31 October
(MJD 58787), where X-ray bursts are observed in ObsIDs
2200840123–25. The detailed analysis of the spectral and
variability evolution of the full outburst will be presented
elsewhere.
To process the data, we used HEASOFT v6.26.1 and
NICERDAS v6 (HEASARC 2014), and applied standard filter-
ing criteria, i.e., pointing offset < 54
00
, Earth limb elevation
angle > 15
◦
, and bright Earth limb angle > 30
◦
. Because
the observations occurred during an epoch of high optical
loading, the parameter “underonly_range” was increased to
0 − 400 when reprocessing the data using the nicerl2 tool.
The first X-ray burst occurred during a nominal South At-
lantic Anomaly (SAA) passage. As such, “nicersaafilt”
was set to “no” on the second observation analysed, in order
to include the data taken during the SAA passage. To test if
the data were affected by the apparent passage through the
SAA, we extracted the 13 −15 keV lightcurve to look for the
presence of high-energy background flares (see Bult et al.
2018); we found none. The total good exposure after pro-
cessing was 43.6 ksec.
We determined the background contribution of our ob-
servations from NICER observations of a Rossi X-ray Tim-
ing Explorer (RXTE ; Bradt et al. 1993; Jahoda et al. 2006)
blank-sky region (∼ 1–2 cts s
−1
from RXTE -6).
2.1 Light Curves and Hardness Ratios
To study the X-ray bursts in the context of MAXI J1807’s
outburst, we constructed a 0.3–10 keV long-term light curve
using 25 s bins. We extracted individual X-ray burst light
curves in the 0.3–10 keV energy band using 0.1 s bins.
Table 2. The black body temperatures, photon indices, 0.5–
10 keV unabsorbed fluxes, and reduced χ
2
obtained from mod-
elling the pre-burst emissions of MAXI J1807, with all errors
quoted at a 90% confidence interval.
Burst kT
bb
Photon Unabsorbed Flux χ
2
ν
(keV) Index (10
−10
erg cm
−2
s
−1
) (χ
2
/dof)
1 0.10 ± 0.01 2.1 ± 0.2 0.54 ± 0.03 205.8/177
2 0.19 ± 0.01 2.0 ± 0.2 1.75 ± 0.05 263.2/233
3 0.11 ± 0.02 2.6 ± 0.3 1.14 ± 0.04 214.8/178
We define the start of each X-ray burst as the first 0.1 s
bin which has an intensity 1σ above the persistent emission
(calculated as the median count rate of the exposure) when
scanning backwards in time from the X-ray burst peak. We
define the end of an X-ray burst as the initial time of a 10 s
bin that is consistent within 1σ of the median count rate
of the last 100 s of the data segment containing the X-ray
burst.
The hardness ratio was defined as the count rate in the
1–10 keV energy band divided by the count rate in the 0.3–
1 keV energy band. Light curves for the hardness ratios were
extracted with 0.2 s binning.
2.2 Spectral Analysis
2.2.1 Persistent Emission
Before examining the X-ray burst emissions, the pre-burst
(persistent) spectra of MAXI J1807 were first explored. For
this, we extracted the spectra of the persistent emission from
exposures of 509 s, 395 s, and 110 s before the first, second,
and third X-ray bursts, respectively. The 0.5–10 keV energy
spectrum was then successfully fitted with an absorbed black
body plus power-law model in Xspec v12.10.1:
tbabs (bbodyrad + powerlaw) ,
where tbabs takes into account the effect of the interstellar
absorption (Wilms et al. 2000). This simple model fits the
data well. As the burst flux is generally much brighter than
the persistent emission, our results remain unaffected if we
use more complex models for the persistent emission. We
found a column density of N
H
= (1.3 ± 0.9) × 10
21
cm
−2
before the first X-ray burst and chose the same value to fit
the persistent emission spectra before the second and third
X-ray bursts. We note that our results were the same within
error by (i) using the column density as explained above,
(ii) using column densities as estimated from the persistent
emission before each burst, or (iii) using an average of the
three column densities.
The black body temperatures, photon indices, 0.5–
10 keV unabsorbed fluxes, and reduced χ
2
obtained from
modelling the pre-burst emissions are reported in Table 2.
In all cases, we obtained a reduced χ
2
of ∼ 1.1. The errors
are quoted for a 90% confidence interval.
2.2.2 Time-Resolved Spectroscopy
We performed time-resolved spectroscopy on all three X-ray
bursts. Each X-ray burst was divided into time bins con-
taining a minimum of 500 counts. We fitted energy spectra
Downloaded from https://academic.oup.com/mnras/advance-article/doi/10.1093/mnras/staa3657/6000262 by DTU Library user on 26 November 2020
ORIGINAL UNEDITED MANUSCRIPT
4 A. C. Albayati et al.
to each time bin using the variable persistent flux method
(f
a
-method; see, e.g., Worpel et al. 2013, 2015).
The f
a
-method describes the X-ray burst thermal emis-
sion with a black body component and any possible excess is
accounted for by scaling the pre-burst emission. This model
is usually constructed by first fixing the pre-burst spectral
components and then adding a black body component for
the X-ray burst emission. A multiplicative factor f
a
is then
applied on the persistent part, which allows us to account for
the effect of X-ray burst emission on the accretion processes.
In Xspec, this model reads as:
tbabs ( bbodyrad + f
a
(bbodyrad + powerlaw) ) ,
where tbabs takes into account the interstellar absorption
and bbodyrad the black body emission from the X-ray
burst. In our study, the model for the persistent emission
(bbodyrad + powerlaw) is fixed to the respective values we
found in Section 2.2.1.
The f
a
-method has already been used for time-resolved
spectroscopy of X-ray bursts using NICER data (see, e.g.,
Keek et al. 2018; Jaisawal et al. 2019). Although the f
a
-
method has been shown to be a good way to parame-
terise the effects on the accretion disc, it is likely that f
a
should be a function of energy rather than a constant (see,
e.g., Keek et al. 2014; Degenaar et al. 2018). In our analysis,
given the relatively low count rates during MAXI J1807’s
X-ray bursts, we assumed an f
a
that does not depend on
energy.
3 RESULTS AND DISCUSSION
3.1 X-ray Imaging Observations
NICER is a non-imaging instrument. Here we investigate
whether the X-ray bursts could originate from a different
source than MAXI J1807.
MAXI J1807 was observed with NuSTAR
(Harrison et al. 2013) on 26 September 2019, i.e., about
35 days before the first X-ray burst we detected with
NICER. In Figure 1 we show the 3–78 keV NuSTAR
image including the NICER field-of-view (a circle with
3 arcmin radius, corresponding to a 30 arcmin
2
field-of-view;
Gendreau & Arzoumanian 2017). Only one source was
significantly detected, with coordinates consistent with that
of MAXI J1807.
There are also 28 photon counting (PC) mode
Swift/XRT observations taken over the period 26 March–
30 May 2017 (there are no Swift/XRT imaging observations
during 2019). We created a 0.3–10 keV image of the NICER
field-of-view by integrating the 28 images (total exposure
time of 24 ks). The image was reduced using the Swift/XRT
data product generator provided by the University of Leices-
ter
1
. Only one source was significantly detected, with coor-
dinates consistent with that of MAXI J1807. Given the evi-
dence provided by NuSTAR and Swift/XRT, we concluded
that the X-ray bursts originate from MAXI J1807.
1
https://www.swift.ac.uk/user_objects/index.php
0 1 2 3 4
Time since MJD 58783 (days)
0
25
50
75
100
125
150
175
NICER 0.3-10keV Count Rate (cts s
−1
)
Figure 2. NICER long-term light curve of MAXI J1807 with 25 s
time resolution in the 0.3–10 keV energy band. This section of the
outburst contains the three thermonuclear X-ray bursts detected
in October 2019. The X-ray burst data have been removed for
clarity, and their onsets marked by arrows.
3.2 Outburst Evolution and Occurrence of X-ray
bursts
Figure 2 shows the outburst evolution of MAXI J1807 from
27 to 31 October 2019 (MJD 58783–58787). The occurrence
of the X-ray bursts are marked by arrows. On 27 October
(t ≈ 0.7 days), the light curve shows a flare-like feature,
where the count rate rises from ∼ 2 cts s
−1
to ∼ 15 cts s
−1
and falls back to the original count rate within half a day.
Roughly 3 hrs later, we observe the onset of a larger flare
with multiple peaks. The count rate increases to ∼ 70 cts s
−1
over approximately 3 hrs, after which the first X-ray burst
occurs on MJD 58784.61727 (t ≈ 1.6 days in Figure 2). The
flare reaches a first peak at ∼ 110 cts s
−1
roughly 4.5 hrs af-
ter the X-ray burst, and drops to ∼ 95 cts s
−1
in the fol-
lowing 4 hrs. There is an additional peak almost 10 hrs
later during which the second X-ray burst occurs on MJD
58785.50558 (t ≈ 2.5 days in Figure 2) at a persistent count
rate of ∼ 150 cts s
−1
. The persistent count rate then drops
to ∼ 80 cts s
−1
over approximately 10 hrs before a final rise
to ∼ 90 cts s
−1
, after which the flare begins to decay. At this
stage in the outburst the third X-ray burst occurs on MJD
58786.48803 (t ≈ 3.5 days in Figure 2) at a persistent count
rate of ∼ 80 cts s
−1
. There is a sudden decrease in flux just
less than a day after the third X-ray burst, where we ob-
serve the flux decreasing from ∼ 60 cts s
−1
to ∼ 10 cts s
−1
in
roughly 4 hrs.
We observed the three X-ray bursts over a period of
three days. The waiting time between the first and second
X-ray bursts was 21.3 hrs, while between the second and
third X-ray bursts was 23.6 hrs. We note that the data-gaps
prevent us to understand whether there were other X-ray
bursts than those we detected.
3.3 Type-I X-ray Burst Light Curves
Figure 3 shows light curves and hardness ratios of the three
X-ray bursts exhibited by MAXI J1807. The first, second,
and third X-ray bursts will hereafter be referred to as B1,
B2, and B3, respectively. At the burst onset, t
B1
= 0, B1
count rate increases from the persistent rate of ∼ 62 cts s
−1
Downloaded from https://academic.oup.com/mnras/advance-article/doi/10.1093/mnras/staa3657/6000262 by DTU Library user on 26 November 2020
ORIGINAL UNEDITED MANUSCRIPT
Type-I X-ray Bursts from MAXI J1807+132 5
0
5
10
15
20
25
Count Rate (x10
2
cts s
−1
)
1a
p.e. 62 cts s
−1
2a
p.e. 148 cts s
−1
3a
p.e. 83 cts s
−1
0 10 20 30
0
2
4
6
Hardness Ratio
1b
0 10 20 30
Time (s)
2b
0 10 20 30
3b
2 4
0
5
10
15
0 2
0
2
4
6
Figure 3. Top panels: light curves of the three X-ray bursts from MAXI J1807 with 0.1 s time resolution in the 0.3–10 keV energy
band. We found the persistent emission levels by taking the mean count rate for 100 s of data before the start of each X-ray burst, and
subtracted them from the light curves. The persistent emission count rates are denoted on the panels as “p.e.”. The insets in panels 1a
and 3a show rise features in detail, which are also shaded in the main panels. Bottom panels: hardness ratios (1–10 keV / 0.3–1 keV count
rate) with 0.2 s time resolution.
to ∼ 500 cts s
−1
over approximately 2 s. At this point, the
burst exhibits a “pause” lasting ∼ 1.6 s (shaded region and
inset in Figure 3, panel 1a). Following the pause, the burst
reaches its peak in roughly 0.2 s. The count rate remains
approximately constant during this peak at ∼ 2250 cts s
−1
between t
B1
= 3.8 − 5.1 s. After the peak, the count rate de-
creases for roughly 133 s to a persistent rate of ∼ 70 cts s
−1
,
i.e., at a flux ∼ 13% higher than the persistent emission be-
fore the burst onset. The overall burst duration is ∼ 137 s.
The rise of B2 is in two parts. In the initial rise, starting
at t
B2
= 0, the count rate increases from the persistent rate
of ∼ 148 cts s
−1
to ∼ 460 cts s
−1
over approximately 2 s. The
count rate then rises suddenly and reaches the peak of the
burst in less than one second. This peak has an average
count rate of ∼ 920 cts s
−1
between t
B2
= 2.6 − 4.6 s. As the
data segment containing the burst ends less than 100 s after
the burst peak, the end of the burst is ill-defined while the
count rate is still 8% above the pre-burst level (160 cts s
−1
,
the median count rate of the last 20 s of the data segment
containing B2).
B3 exhibits interesting features in both its rise and
decay. At t
B3
= 0, B3’s count rate increases for roughly
1.5 s from the persistent rate of ∼ 83 cts s
−1
to a potential
pause similar to that we observed in B1, but at ∼ 290 cts s
−1
(marked by the shaded region and inset in Figure 3, panel
3a). After the potential pause, the count rate continues to
rise reaching a peak average count rate of ∼ 1417 cts s
−1
be-
tween t
B3
= 4.0 − 5.3 s. During the decay there is indica-
tion of a double peak at t ≈ 7.5 s lasting roughly 2 s; how-
ever the double peak is not statistically significant given the
large error bars. After this, the count rate decreases over
approximately 78 s to a post-burst persistent count rate of
∼ 100 cts s
−1
, which is ∼ 20% higher than the pre-burst per-
sistent rate. The overall burst duration is ∼ 83 s.
The hardness ratios of all three X-ray bursts track sim-
ilar profiles to the light curves, increasing through the burst
rise and decreasing through the decay. The hard band (1–
10 keV) dominates after less than 1 s in each case, and in-
creases to peaks of ∼ 4 in B1 and B3, and ∼ 3 in B2. There is
a plateau in the hardness ratio that starts during the pause
in B1 and appears to continue after it. There is also evidence
of a plateau during the potential pause found in B3, after
which the ratio increases; however in this case the plateau
is not as well constrained as in B1.
We measure different average count rates for the per-
sistent emission before and after each X-ray burst. Given
our data set, we are unable to understand if this is due to
an intrinsic change in the flux of the persistent emission, or
whether it is the effect of long decay tails which we cannot
differentiate from the continuum (see, e.g., in’t Zand et al.
2017). Out of the three X-ray bursts, B1 has the highest
peak count rate, whilst B2 has the lowest. All three X-ray
bursts have a rise time of ∼ 4 s and exhibit long decay tails
(>1 minute). A slow rise and long decay is indicative of
H-rich fuel at the moment of ignition, which is likely the re-
sult of accretion of a mixed H/He fuel (Galloway et al. 2008;
Galloway & Keek 2017; Schatz et al. 2001).
Downloaded from https://academic.oup.com/mnras/advance-article/doi/10.1093/mnras/staa3657/6000262 by DTU Library user on 26 November 2020
