Astronomy & Astrophysics manuscript no.

ie613 ©ESO 2025

May 15, 2025

Long-term Timing Results of Ecliptic Pulsars Observed with

I-LOFAR
S. C. Susarla 1, O. A. Johnson 2, 3 , D. J. McKenna 4 , E. F. Keane 2 , P. J. McCauley 2, J. P. W. Verbiest 5,

C. Tiburzi 6 , A. Golden 1
(Affiliations can be found after the references)

Received XXX; accepted YYY

arXiv:2505.09549v1 [astro-ph.HE] 14 May 2025

ABSTRACT

Context. Pulsar timing at low frequencies offers a powerful tool for studying the interstellar medium. Additionally, pulsar observations in the
ecliptic enables us to study the effects of the solar wind which becomes much more prominent at low radio frequencies. The Irish station of the
LOw Frequency ARray (I-LOFAR) is a sensitive low-frequency radio telescope, capable of delivering high-precision data for pulsar studies.
Aims. We present a comprehensive dataset of times-of-arrival, timing solutions and dispersion measure (DM) time series for seven ecliptic pulsars
observed over two-to-three years with I-LOFAR. The primary objectives are to investigate time-dependent dispersion effects and provide high-
precision timing data for pulsar timing experiments.
Methods. We measure DM variations through pulsar timing and analysed these across different ecliptic latitudes to assess the impact of the solar
wind on each pulsar. We model the intrinsic pulse-profile variability as a function of frequency.
Results. The high-precision DM time series for all seven pulsars exhibit clear variations dependent on their ecliptic latitudes, revealing the impact
of the solar wind. Some pulsars show significant changes in their pulse widths across the frequency band, while others remain stable. We examine
and quantify the pulse-nulling present in PSR J0826`2637, we report evidence for DM chromaticity in PSR J1645´0317, and we describe how
PSR J2145´0750’s DM precision is such that it could resolve the ionospheric DM contribution. This makes it a target of interest for telescopes in
areas of the globe where the ionospheric electron density is higher, e.g. the Murchison Radio Observatory in Australia.
Conclusions. This data release underscores the potential of I-LOFAR, or any standalone international LOFAR station, for advancing low-frequency
pulsar studies, particularly in analyses of dispersion in the interstellar medium, the solar wind and the ionosphere.
Key words. radio astronomy – pulsar timing methods – solar wind – ionosphere – methods:data analysis – pulsars:general

1. Introduction tematically identified by examining correlated patterns in the

residuals across an array of pulsars (Hellings & Downs 1983);
Pulsars are highly magnetized, fast-spinning neutron stars, this approach underlies the concept of pulsar timing array (PTA)
remnants of massive stars following a supernova (Gold 1968; experiments. The current state-of-the-art (Verbiest et al. 2024)
Lorimer & Kramer 2004). Pulsars emit beams of electromag- has seen hints of such a correlated signature in data from vari-
netic radiation from their polar regions (Contopoulos et al. 1999; ous PTA consortia at a significance level from 3 to 4.2σ (Anto-
Michel & Li 1999; Philippov et al. 2015); and as their mag- niadis et al. 2023a; Agazie et al. 2023; Reardon et al. 2023; Xu
netic and rotational axes are in general misaligned, these beams et al. 2023; Agazie et al. 2024). In identifying any GW signal
sweep across space while the star rotates. When one of the beams in PTA datasets, one of the most challenging signatures to dis-
cuts through the line-of-sight (LoS) of an observer from Earth, entangle is that of the ionised interstellar medium (IISM). The
regular pulses of radiation are observed, which are particularly IISM noise process can dominate the GW background if not
prominent at radio frequencies. These pulses, when folded over modelled properly, but its inverse dependency on the squared
the rotational period, form a characteristic pulse profile that is of the observing frequency offers a leverage to implement ef-
stable over time (Helfand et al. 1975; Liu et al. 2012). This fective mitigation techniques, especially thanks to the usage of
makes pulsars ideal for a number of experiments, based on the low radio-frequency pulsar data. Indeed, one of the best ways to
technique of pulsar timing (see e.g. Stairs 2003 and references characterise the IISM noise process is to observe pulsars with
therein). Pulsar timing is a process that entails the recording of facilities that reach radio-frequencies ă 1 GHz, and large frac-
the times-of-arrival (ToAs) of pulses emitted from a given pul- tional bandwidths (Verbiest & Shaifullah 2018).
sar (Edwards et al. 2006). ToAs are derived by cross-correlating
a pulsar signal integrated over many rotation periods, typically This paper presents a study of seven pulsars observed be-
104 to 105 , with a high signal-to-noise (S/N) template (Tay- tween 100 and 200 MHz, with high fractional bandwidth over
lor 1992). Given the stability requirement, “high-precision pul- the last two-to-three years, using the Irish station of the Low Fre-
sar timing” (with a residual root-mean-square, rms, lower than quency Array (LOFAR; see van Haarlem et al. 2013), hereafter
1µs) is achieved by long-term monitoring of millisecond pul- referred to as I-LOFAR (also known as IE613). These seven pul-
sars (MSPs; see Backer et al. 1982), rather than the majority of sars are located near the ecliptic plane with various ecliptic lat-
pulsars which rotate with periods of order of half a second or itudes (ELAT), so that their signal traverses the solar wind once
more. Several phenomena introduce noise into the timing model per year during their Solar conjunctions, offering an excellent
(see Verbiest & Shaifullah (2018) for a review). Among these means of assessing the solar contribution to the IISM signature.
noise sources are gravitational waves (GWs), which can be sys- The data products developed as part of this work are provided
Article number, page 1 of 11
 A&A proofs: manuscript no. ie613

open source (Susarla 2025)1 . This paper describes these data pulsars listed in Tiburzi et al. (2021) at least once to determine
and our pipelines; and illustrates some of the enabled scientific optimal integration times while adhering to the telescope’s op-
investigations. In § 2 we outline the dataset and our data col- erational limit of 60 minutes per observation. Given the weekly
lection methods, including choices made for the observational constraint of a maximum observing time of 31 hours, we selected
settings and the initial processing steps. In § 3 we describe the the seven best sources, optimizing the sample based on available
post-processing techniques applied to the data. § 4 presents the time and ensuring a spread in Right Ascensions. Out of the seven
results and discusses their implications; finally, § 5 concludes selected sources, two are part of PTA samples.
with a summary of the findings and future research directions. Fig. 1 displays the location of these pulsars relative to the
Sun’s path over a year. Notably, each pulsar’s LoS aligns closely
with the Sun at different times of the year, depending on its Right
2. I-LOFAR: Telescope and Observing campaign
Ascension. Tab. 1 provides a summary of key properties for
In this section, we describe the Irish international LOFAR these pulsars, including the median signal-to-noise ratio (S/N),
(I-LOFAR) telescope used for this study and its configurations which indicates their relative brightness in the LOFAR (HBA)
for pulsar data collection. We also describe the selection criteria frequency range.
that led to the choice of the observed pulsars.
2.3. Dispersion Measure precision
2.1. I-LOFAR Telescope As an electromagnetic wave traverses the IISM, it undergoes
LOFAR is a pan-European radio telescope array consisting dispersion due to the free electrons present in it. This effect is
of 52 stations from as far West as Ireland to as far East as Latvia. characterised by the dispersion measure (DM) parameter, de-
The core of the array is in the Netherlands where there are 24 fined as the total electron density along the line of sight between
core stations centred on a “superterp” near Exloo, the Nether- Earth and a pulsar, affecting the arrival times of radio pulses
lands. There are a further 14 remote stations within the Nether- by causing frequency-dependent delays. The DM of a pulsar is
lands and 14 international stations. At present the whole ar- given by the integral of the column density neLoS along the line of
ray is being upgraded (LOFAR Collaboration 2023) to, amongst sight (LoS):
other improvements, increase the instantaneous bandwidth and ż
the clock distribution. Furthermore two new international sta- DM “ neLoS dl. (1)
tions, in Italy and in Bulgaria, will soon join the array, signifi- LoS
cantly improving its North-South extent. The dataset collected The dispersion causes a delay in the arrival of pulsar emission
with the Irish LOFAR station, is the subject of this paper. I- that depends on both the frequency of the radiation and the DM
LOFAR is located at the Rosse Observatory2 in Birr, Co. Of- parameter, expressed in units of pc cm´3 . This delay can be cal-
faly, in the centre of the island of Ireland. During the observ- culated as:
ing campaign, I-LOFAR operated in standalone mode for typi-
DM
cally 31 hours per week and it was within this allocation, over ∆t “ , (2)
the past three years that data were collected for the presented K D ν2
project. The I-LOFAR high-band antennas (HBA) were used, where KD is the dispersion constant, and ν is the observing fre-
observing between 100 and 200 MHz. For each observation the quency in MHz. The constant KD itself is a combination of
HBA were analogue-beamformed in the direction of the target physical constants, each measured to some finite level, and SI
with the resultant data sampled using a 200-MHz clock, result- conventions. We note that in our work here we adopt a stan-
ing in one real data sample per antenna polarisation every 5 ns. A dard pulsar timing convention to fix this constant at the exact
coarse channelisation is performed using a polyphase filterbank value of KD B 2.41 ˆ 10´4 MHz´2 pc cm´3 s´1 (Lorimer &
which creates 512 channels of complex voltage data per polari- Kramer 2004). This is consistent with the relevant pulsar soft-
sation every 5.12 µs. The hardware limits the accessible data to ware packages, namely, DSPSR (van Straten & Bailes 2011)
488 of these 512 sub-bands for 8-bit recording. The data in the and tempo2 (Hobbs et al. 2006). As the dispersion delay is in-
vicinity of the FM band contain no useful information due to a versely proportional to the square of the frequency, it becomes
hardware filter, thus we choose to include in the final observa- significantly more pronounced at low frequencies. In this study,
tion sub-bands 12–499, that correspond to a lowest and highest the average DM precision obtained on an individual observation
topocentric frequencies of about 102.2461 and 197.5586 MHz. across all pulsars is 1.2 ˆ 10´4 pc cm´3 .

2.2. Pulsar selection 2.4. Observations and data analysis

Since its inception, I-LOFAR has been conducting pulsar ob- During the standalone time, regular observations of the seven
servations. For this study, we focused on a carefully selected chosen pulsars were conducted over the course of two-to-three
sample of seven pulsars observed consistently over a two-to- years, accumulating a total of „240 hours of combined obser-
three year period, each satisfying the condition |ELAT| ă 20˝ . vations. Science-ready data products were produced using the
These pulsars were chosen based on the list published in Tiburzi REAL-time Transient Acquisition backend (REALTA; Murphy
et al. (2021), ensuring a wide temporal coverage of solar con- et al. 2021). Baseband voltages were recorded to disk, and udp-
junctions throughout the year. Our aim was twofold: to enhance PacketManager (McKenna et al. 2023) was used to create in-
the cadence of the low-frequency dataset contributed to the In- termediate data products in the PSRDADA3 format from these
ternational PTA (IPTA) database (Perera et al. 2019) for the PTA voltages. Furthermore, coherent dedispersion is applied to the
pulsars and to ensure a robust data coverage for studying so- baseband voltages using the digifil tool of DSPSR. Coher-
lar wind (SW) effects. To finalize the sample, we observed all ent dedispersion is a technique wherein an input signal is con-
1
volved with the inverse transfer function of the IISM, effectively
https://doi.org/10.5281/zenodo.14865192
2 3
https://www.rosseobservatory.ie/ https://psrdada.sourceforge.net/index.shtml

Article number, page 2 of 11

 LOFAR: Long-term Timing Results of Ecliptic Pulsars Observed with I-LOFAR

Fig. 1: An Aitoff projection of the locations of the pulsars listed in Tab. 1. The pulsars are shown in star symbols as well as the
ecliptic plane, which is indicated by a yellow dashed line.

Pulsar Name Time span, tspan Average Integration Period Ecl. Lat. DM Median
(J2000) (MJD) (yr) time (min) (ms) (deg) (pc/cm3 ) S/N
J0034´0534 59892´60431 (2.0 yr) 50 1.8 ´8.53 13.7650 72.2
J0826+2637 59695´60411 (2.5 yr) 15 530.6 7.24 19.5041 522.5
J1022+1001 59803´60542 (2.0 yr) 50 16.5 ´0.06 10.2532 40.6
J1136+1551 59695´60430 (2.5 yr) 15 1187.9 12.17 4.8482 917.7
J1300+1240 59695´60431 (2.5 yr) 60 6.2 17.63 10.1588 39.5
J1645´0317 59696´60431 (2.5 yr) 20 387.6 18.89 35.7416 612.0
J2145´0750 59696´60530 (3.0 yr) 45 16.1 5.31 9.0027 119.8

Table 1: Key properties of the observed ecliptic pulsars presented in this study. The table summarizes the pulsars’ J2000 names,
observation timespans in MJD and years, average integration times, rotational periods, ecliptic latitudes, fiducial DMs, and the
median S/Ns over the entire observing span.

removing the dispersion delay signature caused by free elec- and partially frequency-averaged to ten frequency sub-bands us-
trons along the line of sight (Hankins & Rickett 1975). Dual- ing the PSRCHIVE software suite (Hotan et al. 2004; van Straten
polarisation complex products are produced for each observa- et al. 2012). The final bandwidth was cut to range from 112 to
tion, which were then converted into SigProc (Lorimer 2011) 190 MHz, as a small portion of the wider band was affected by
filterbank files using the digifil tool with a frequency resolu- consistent RFI and the edges of the high/low pass filters present
tion of 24.41 kHz and source-dependent temporal resolution of in the system.
at least 81.92 µs. The filterbanks are then converted into TIMER
archive (Hotan et al. 2004) files, a standard pulsar data format,
by folding them using dspsr. 3.1. Creation of templates

3. Data processing All time-averaged archives, except those within 45 de-

grees of the Sun, are combined using the psradd tool
The final data have a central frequency of 149.902 MHz and
from PSRCHIVE to make one individual archive file. For
were coherently dedispersed to this frequency. Then, they were
PSR J0034´0534, given its complexity, we selected the highest
folded into 10 s sub-integrations, using the DSPSR software
S/N observation instead of adding all the observations. We then
suite. Once the archives were created according to these specifi-
frequency-average the channels down to 10, and smoothened
cations, several post-processing steps were applied. In the post-
such data-derived templates with the wavelet-smoothing tech-
processing phase, each observation was cleaned of radio fre-
nique offered by the psrsmooth tool. This process is performed
quency interference (RFI) using a modified version of the Coast-
iteratively after optimizing the pulsar model (or ephemeris)
Guard software package (see Lazarus et al. (2016), Kuenkel, in
via the timing technique (see § 3.2), updating the archives’
prep.4 ) and corrected for parallactic angle rotation and projection
ephemerides to ensure accurate DM values in the profiles. The
effects (Jones 1941; Hamaker et al. 1996) with the dreamBeam
presence of any residual dispersion in the final template is
software package5 . Each observation was then time-averaged
checked using the pazi module, with corrections applied to the
4
https://github.com/larskuenkel/iterative_cleaner DM as needed. The result is a high-S/N, noiseless template (see
5
https://github.com/2baOrNot2ba/dreamBeam Fig. 2).
Article number, page 3 of 11
 A&A proofs: manuscript no. ie613

Fig. 2: Frequency evolution of normalized pulse profiles. The colour intensity is scaled according to the observing frequency. Darker
shades correspond to a high frequency and lighter shades correspond to a lower frequency.

3.2. Pulsar timing binary pulsars. This fitting process is typically repeated, until a
stable timing solution is obtained. The two PTA sources in our
The procedure described in the previous section is used to sample, PSRs J1022+1001 and J2145´0750, are usually pro-
create a preliminary template from all of the cleaned archives. vided with an already, relatively stable timing solution thanks
ToAs were then created for each observation (that was time- to the European PTA Data Release 2 (Antoniadis et al. 2023b).
averaged and frequency-averaged down to 10 channels) and Once, a reliable timing solution was obtained, the new ephemeris
channel via cross-correlation with the template. This ensemble file was installed in the original observations, that are then time-
of ToAs was used to generate timing residuals and to fit a stable averaged again. These archives were again used to make the final
timing solution. tempo2 is utilized to fit for various timing pa- template repeating the procedure mentioned in § 3.1.
rameters, including position, proper motion, spin frequency and
its derivatives, DM and its derivatives, and binary parameters for
Article number, page 4 of 11
 LOFAR: Long-term Timing Results of Ecliptic Pulsars Observed with I-LOFAR

3.3. Obtaining DM time-series

Using the noise-free template generated by the methods in
§ 3.1, the pat module was employed to produce topocentric
ToAs for each frequency channel. Following Tiburzi et al. (2019)
and Donner et al. (2020), a DM time-series was generated us-
ing an approach known as the “Epoch-wise” method (Iraci et al.
2024). This technique fits the ToAs belonging to the same ob-
serving epoch to the functional form shown in Eq. 2 using
tempo2 (Hobbs et al. 2006), to obtain the residual DM induced
by the added time delay. The result is a DM measurement per
epoch. All data products—including the DM time series, tim-
ing solutions, and ToAs—were obtained concurrently using the
methods described in this section. The resulting spin and orbital
parameters for all pulsars are summarised in Table 3.

4. Results and Discussions

This section outlines the properties of each pulsar. As already
mentioned, PSRs J1022+1011 and J2145´0750 are included in
the samples of the European and International PTAs; they are
old, recycled pulsars with periods of less than 20 ms. Their sta-
bility and predictability make them ideal candidates for precision
pulsar timing and for gravitational wave studies (Verbiest et al.
2016; Perera et al. 2019).

Pulsar Name DM Uncertainty (ˆ10´5 pc cm´3 )

(J2000) Median Minimum
J0034´0534 6 1 Fig. 3: DM time series of each of the pulsars. The red points
J0826+2637 19 5 in each panel are the epochs that are less than 45 degrees away
J1022+1001 16 2 from the Sun.
J1136+1551 13 3
J1300+1240 13 4
J1645´0317 9 3 using Eq. 22 in Bray et al. (2015), we get a DM contribution of
J2145´0750 8 0.9 3 ˆ 10´6 ´ 1.9 ˆ 10´5 pc cm´3 . The high fractional bandwidth
and enhanced sensitivity of future instruments like the Square
Table 2: DM uncertainties for selected pulsars observed with I- Kilometer Array (SKA) (Braun et al. 2019) will enable signifi-
LOFAR, showing the median and minimum uncertainties. cantly improved precision, providing a promising avenue for de-
tailed investigations of the ionosphere using pulsars.

4.2. Frequency Evolution of Pulsar Pulse Profiles

4.1. DM timeseries
Multi-frequency pulse profiles offer crucial information on
The DM time series are displayed in Fig. 3, with red points pulsar emission geometries, including emission region heights,
marking epochs when the pulsar’s LoS comes within 45˝ of the beam shapes, and radius-to-frequency mapping (Lyne & Manch-
Sun. Significant fluctuations in the DM time series are observed ester 1988; Rankin 1983a,b; Mitra & Deshpande 1999; Hankins
especially for PSRs J0034´0534, J0826+2637, J1022+1001, & Rankin 2010; Hassall et al. 2013). A comprehensive study of
and J2145´0750 as their LoS near the Sun. These emphasize 74 pulsars by Xu et al. (2021) provided significant information
their potential for studying SW effects, as their ecliptic latitudes on the frequency evolution of pulse profiles. For 71 pulsars, the
make them ideal candidates for such investigations using pulsars. pulse width W50 (full width of the pulse profile at 50% of the
Other pulsars, such as PSRs J1300+1240 and J1645´0317, peak profile amplitude, FWHM), follows a power-law trend at
show gradual trends in their DM time-series, indicating poten- both high and low frequencies. However, deviations from this
tial variations in the IISM electron content along their line of power-law behaviour were observed in the mid-frequency range
sight. These gradual changes may reflect evolving electron den- (100 to 800 MHz), consistent with absorption characteristics
sities within the IISM, highlighting the importance of studying proposed for the first time by Rankin (1983b).
long-term dispersion effects with low-frequency observations. Following Xu et al. (2021) we use the FWHM as a metric to
Tab. 2 provides details of each pulsar along with their DM un- study the beam widths as a function of frequency of the pulsars
certainty. Some pulsars demonstrate exceptionally high DM pre- in our sample at the exceptionally long wavelengths offered by
cision, suggesting that our observations are approaching the sen- I-LOFAR. We obtain W50 by fitting multiple Gaussians, i.e.
sitivity needed to detect ionospheric variations within the DM n
˜ ¸
time-series. If we consider a total electron content of the iono- ÿ ´px ´ µi q2
y“ Ai exp , (3)
sphere to be 10-60 TECU which is the normal value over a full
i“1
2σ2i
day at Birr (see Kriegel & Berdermann (2020)6 for TECU maps),
where Ai , µi and σi are amplitude, mean and standard deviation
6
https://impc.dlr.de of the ith Gaussian component. The W50 and its uncertainty W50,i
err

Article number, page 5 of 11

 A&A proofs: manuscript no. ie613

Fig. 4: Full-width-half-maximum (FWHM) pulse widths for pulsars in our dataset. The black stars represent the pulse width at each
frequency with the corresponding error in measurement. The red lines in each panel are the power law fits on the FWHM values.
The FWHM values that lie below the power law fits correspond to the absorption features. It can be seem most prominently in
PSRs J1300+1240 & J1645´0317.

are then given by: why the W50 values for PSR J1136+1551 appear to be quantised.
? To get the error on the FWHM, we used the quadrature sum of
W50,i “ 2 2 ln 2 σi , err
the W50,i of the individual Gaussian components. The frequency
? evolution of these pulsars is fit with a power law function of the
err
W50,i “ 2 2 ln 2 σerr (4)
i form W50 9 f α .
where σerri is the uncertainty on the standard deviation of the Fig. 4 shows W50 as a function of observing frequency. The
Gaussian component. For pulsars with multiple peaks exceed- red lines represent the power law fits on the W50 values. We see
ing 50% of the peak amplitude, where the intervening trough that for some pulsars the W50 values lie below the power law
falls below this threshold, such as certain frequency profiles of fit. These so-called “absorption” features, could pose challenges
PSR J0034´0534 and all profiles in PSR J1136+1551, we man- to the conventional pulsar radiation models (Rankin 1983b; Xu
ually measured W50 as the individual W50,i values or their sum et al. 2021). They are attributed to cyclotron absorption and
did not accurately represent the beam width. This is the reason it is likely magnetospheric in origin (Bartel 1981). Most pul-
Article number, page 6 of 11
 LOFAR: Long-term Timing Results of Ecliptic Pulsars Observed with I-LOFAR

sars exhibit a general trend of decreasing pulse width with in- 10% of maximum (W10 ) as, depending on the particle profile
creasing frequency. This is the trend expected from a radius- shape (i.e. on the viewing geometry), different width metrics
to-frequency mapping along primarily dipolar field lines where might be more representative of the emission beam widths (Lyne
higher-frequency emission originates closer to the stellar sur- et al. 2010). We saw similar trends across the entire frequency
face (Ruderman & Sutherland 1975). However exceptions exist, range for all sources, with the exception of PSR J2145´0750
notably, PSR J0034-0534 where the W50 values appear to in- where its W10 values seemed to be increasing with frequency,
crease as a function of frequency. Conversely, PSR J1022+1001 contrary to its W50 trend as shown in Fig. 7. This might imply
shows minimal changes in W50 across a wide frequency range, that emission from the centre and edges of the beam are not
indicating stable profile morphology. PSR J0826+2637 agrees coming from the same height above the surface for the same
with the findings of Rankin (1983b) and Xu et al. (2021) , frequency, i.e. radius to frequency mapping does not hold.
with a steady decrease in their pulse widths over the observed These findings underscore the importance of having a high
frequency range showing no noticeable absorption features. fractional bandwidth. SKA1-Low, which is currently under con-
Whereas, PSRs J1300+1240, J1645´0317 and J2145´0750 struction in Western Australia, has a 7:1 instantaneous frac-
show absorption features in the LOFAR band itself. To address tional bandwidth (Braun et al. 2019) providing the coverage
these exceptions, we also analysed the values of full width at that should enable comprehensive measurements of these trends

Fig. 5: Residuals after timing using tempo2. The different colours are representative of different frequencies.

Article number, page 7 of 11

 A&A proofs: manuscript no. ie613

4.3. PSR J0034´0534

PSR J0034´0534, classified as a binary millisecond pul-
sar (Bailes et al. 1994), is one of the most precisely
timed pulsars at low frequencies. We report a DM of
13.764996 ˘ 0.000015 pc cm´3 with a remarkable median un-
certainty of 7.9 ˆ 10´5 pc cm´3 for all observations. This high
precision in DM measurements allows it to serve as an excellent
probe for detecting subtle DM variations due to effects such as
the SW and the ionosphere. With a rotational period of 1.8 ms,
J0034´0534 has the third shortest period known among pulsars.
It orbits a white dwarf companion in a 1.6 day binary system.
It has a broad integrated pulse profile with a minimal off-
pulse region at 150 MHz. This can be observed in the template
profile shown in Fig. 2. The timing residuals have an rms un-
certainty of 18.54 µs as shown in Fig. 5. Notably, for this pul-
sar the upper three frequency channels were excluded from the
analysis due to the complete absence of signal. This manual
procedure was required, likely as a result of the source’s very
steep spectrum. Additionally, the exceptional DM precision of
J0034´0534 provides evidence for potential asymmetry in SW-
Fig. 7: Full width at 10% of the peak flux density (W10 ) for induced DM variations around solar conjunction, as noted in pre-
PSR J2145´0750 as a function of frequency. Unlike the W50 vious studies (Tiburzi et al. 2019; Susarla et al. 2024). Further
values shown in Fig. 4, W10 exhibits a clear increasing trend. investigation, ideally involving daily observations around solar
The red line indicates a power-law fit to the data. conjunction, would be valuable to determine the sensitivity of
this pulsar to such effects.

4.4. PSR J0826+2637 (B0826+23)

One of the earliest discovered pulsars, PSR J0826+2637,
in the Southern pulsar population and provide unique con- was first identified by Craft et al. (1968) with a rotation pe-
straints on pulsar radiation models. With the aforementioned riod of 530 ms. Using the I-LOFAR dataset, we report a DM of
observed behaviour, and as a very bright source in this band, 19.504089 ˘ 0.00009 pc cm´3 with an rms error in timing resid-
PSR J2145´0750 is an obvious target for SKA1-Low even in uals of 101.40 µs. Although it is among the brightest known pul-
the very early array assemblies. sars, it exhibits complex emission characteristics that have been

Fig. 6: The observation of PSR J0826+2637 from the observation on April 18th, 2023. The left column shows the emission as a
function of time in 10-s sub-integrations; the transition in emission modes is evident. The middle column shows the pulse profile
during the B (Q) mode in the top (bottom) panel. The right column shows the ‘Joy-Division’ plot, displaying rotation period numbers
650 to 800. The transition from the Q-mode to the B-mode is marked by a sudden increase in pulse brightness.

Article number, page 8 of 11

 LOFAR: Long-term Timing Results of Ecliptic Pulsars Observed with I-LOFAR

investigated extensively in studies such as Sobey et al. (2015) 4.6. PSR J1136+1551
and Rankin et al. (2020). PSR J1136+1551 (or B1133+16) is an isolated pulsar with
This pulsar displays two primary emission states: the Bright, a period of 1.18 s, characterized by a double-peaked pro-
or "B" mode, and the Quiet, or "Q" mode. The Q mode is char- file and a high S/N. In our analysis, we report a DM of
acterized by weak and sporadic emission. Additionally, Sobey 4.84821 ˘ 0.00007 pc cm´3 , with an rms timing residual er-
et al. (2015) suggests the possible existence of a third, precursor ror of 122.38 µs. Previous studies have documented the occur-
mode with even lower flux that occasionally drops to zero, poten- rence of unusually intense single pulses, known as giant pulses
tially reflecting magnetospheric processes; however, confirming (Kazantsev & Potapov 2015), which provide an opportunity to
this hypothesis requires an extended observation window. explore the pulsar’s emission mechanisms in greater depth (Os-
Fig 6 shows the two modes observed during the observa- wald et al. 2019).
tion on 18th April 2023. The pulsar appears to be very weak in
the Q mode, as evidenced in many observations where the S/N 4.7. PSR J1300+1240 (B1257+12)
has significantly dropped, in this example by a factor of at least PSR J1300+1240 was the first pulsar found to host a plane-
50. The left column in Fig. 6 shows a waterfall-plot for indi- tary system, with three planets detected in orbit around it (Wol-
vidual pulses on the same date. This precisely shows the pro- szczan & Frail 1992). This discovery was also the earliest evi-
gression from the Q-mode to the B-mode. This poses a chal- dence of extrasolar planets. Located approximately 710 pc from
lenge, albeit tractable, for timing as one must be careful to ac- the Sun, the J1300+1240 system is thought to have originated
count for this moding, i.e. only consider times when the pul- from a white dwarf merger. The presence of orbiting planets
sar is in the B mode. This switching behaviour is seen in the necessitates a complex timing model to account for the influ-
wider pulsar sample when observing sensitivity and cadence al- ences of all planetary bodies. In our analysis, we report a DM
low (Lyne et al. 2010). It appears to be due to magnetospheric of 10.1588 ˘ 0.0028 pc cm´3 , with an rms timing residual error
switching, i.e. the entire pulsar magnetosphere switching from of 30.43 µs. Notably, the DM of this pulsar varies over extended
one stable configuration to another (Timokhin 2010). While this timescales, indicating turbulence in the local ISM. Due to its low
is now widely observed (Keane 2013) it remains an open ques- ecliptic latitude, effects of the SW are seen in I-LOFAR data, par-
tion as to why pulsars switch at all, and why they switch back and ticularly during the first solar conjunction near MJD 59900, as
forth between stable, possibly quantised, states (Cordes 2013). shown in Fig. 3.
This source presents an excellent opportunity to study this phe-
nomenon as it reliably switches in a tractable timescale. It is to 4.8. PSR J1645´0317
be expected that the spin-down rate in each mode is different, as PSR J1645´0317 is a bright pulsar in the LOFAR band
the volume of the magnetosphere and pulsar wind have changed with a period of „387 ms. It has large DM variations with
but the relative time spent in each mode may not permit this to a considerable change of 0.002 pc cm´3 over a period of
be measured for this source. two years (see Fig. 3). In our analysis, we report a DM of
The switch itself is seen here to take less than 1 rotation pe- 35.7416 ˘ 0.0022 pc cm´3 and an rms timing residual error of
riod. Its light cylinder distance is „ 25300? km, „ 84 ms light 57.36 µs.
travel time. The dynamical timescale 1{ Gρ „ 0.5 ms mean- Given the strong DM variations observed in this pulsar, we
ing that such large-scale reconfigurations could conceivably oc- investigated the potential frequency dependence of DM (Cordes
cur this rapidly. Detailed full-polarisation single pulse studies at et al. 2016). To this end, we constructed a standard template
the transitions of this source may be able to pin down the loca- comprising of 20 frequency channels using the methodology
tion and scale sizes of the switches. described in Sect. 3.1 and divided that into two templates of
10 frequency channels each. The observations were frequency-
averaged into 20 channels and subsequently divided into two
sub-bands, each consisting of 10 channels, with central frequen-
4.5. PSR J1022+1001 cies of approximately 130 MHz for the lower sub-band and
170 MHz for the upper sub-band. Following the approach out-
PSR J1022+1001 is another precisely-timed pulsar. Due lined in Sect. 4.1, we generated DM time series for both sub-
to its exceptional stability and relatively low declination, all bands using their respective templates. The resulting DM time
PTAs worldwide include this pulsar in their observations, series, shown in Fig. 8, exhibit systematic differences between
making it one of the most thoroughly studied millisecond DMs measured at the same epoch but in different frequency
pulsars of all times. In our analysis, we report a DM of bands, providing tentative evidence for DM chromaticity. This
10.25321 ˘ 0.00002 pc cm´3 with a median DM uncertainty of effect has been previously reported for PSR J2219+4754 (Don-
1.6 ˆ 10´4 pc cm´3 . Additionally, we report an rms error on the ner et al. 2019) and is primarily attributed to small-scale, high-
timing residuals of 54.92 µs. With an ecliptic latitude of ´0.06˝ , amplitude variations in the interstellar electron density. As dis-
the LoS of this pulsar passes through the Sun. This proximity al- cussed by Cordes et al. (2016), such variations arise from
lows for the examination of SW effects on the pulsar’s signal, as turbulence in the ionised interstellar medium, which induces
previously investigated by studies such as Tiburzi et al. (2021); frequency-dependent multipath propagation due to the result-
Susarla et al. (2024). Particularly, Susarla et al. (2024) showed ing microstructure. While such chromatic DM variations could
that this pulsar has a consistently higher values of electron den- pose challenges for high-precision pulsar timing experiments,
sity at 1 AU after studying a decade’s LOFAR data with a rea- the long-term trends in both sub-bands appear to be consistent.
sonably high cadence. This phenomenon is attributed to the fact Consequently, the impact on timing precision may be negligible.
that the LoS of this pulsar crosses through the slow wind region
of the SW, which is of higher electron density. They also showed 4.9. PSR J2145´0750
the need for a variable model of solar wind for PTA sources like PSR J2145´0750 is a bright pulsar with a period of 16 ms.
these to mitigate this noise effectively. In our analysis, we report a DM of 9.00272 ˘ 0.00001 pc cm´3
Article number, page 9 of 11
 A&A proofs: manuscript no. ie613

is not able to completely model the variability that can be seen

in Fig 5 for pulsars like PSR J1022+1001. The precise DMs ob-
tained in this study emphasize the role of low-frequency obser-
vations in complementing higher-frequency PTA datasets (with
much poorer DM precision), particularly in mitigating DM noise
and SW noise. In gravitational wave studies these noise sources
act as foregrounds (Verbiest & Shaifullah 2018). The precision is
also at the point that it may be possible to detect the variation in
the ionospheric DM using only pulsar timing. The LOFAR DM
precision is „ 30 TECU and other instruments coming online
under more active regions of the ionosphere should have better
resolution still.
This study also examined the frequency evolution of pulse
profiles, shedding light on pulsar emission geometries. While
most pulsars seem to follow a consistent decrease in W50 with in-
creasing frequency, deviations from this, the expected behaviour,
were observed. Separately, we examined the pulse-nulling phe-
nomenon in PSR J0826+2637. This pulsar shows extreme weak-
ness in its Q-mode, with a sudden switch to the B-mode oc-
curring within a single rotation. Such rapid mode transitions
Fig. 8: DM chromaticity in PSR J1645´0317. In the top panel, are likely magnetospheric in origin, suggesting quantized sta-
the black points with errorbars show the DM timeseries at ble magnetospheric states and we see that the switching process
170.6 MHz whereas the orange points show the DM timeseries occurs in less than one rotation period.
at 131.5 MHz. The bottom panel shows the difference in the ab- Furthermore, we investigated the DM chromaticity of
solute values of their DMs at the same epoch. It is worth noting PSR J1645´0317. Our findings present tentative evidence for
that the DM trends in both the bands are different highlighting frequency-dependent dispersion effects in this source. To con-
the frequency-dependent DMs. firm this phenomenon robustly, dedicated observations with ex-
ceptionally high fractional bandwidth and sensitivity at low
frequencies are essential. This could come from a combined
and an rms timing residual error of 28.60 µs. With an ecliptic campaign using, for example, LOFAR and the Giant Metre-
latitude of „ 5˝ , it is an ideal probe to study the effects of SW. wave Radio Telescope, but SKA1-Low’s advanced capabili-
Due to its relatively stable timing solution (Verbiest et al. 2009), ties are probably best suited for such studies, promising sig-
PSR J2145´0750 is also monitored as a PTA pulsar. Its low nificantly improved DM precision. By utilizing bright pulsars
declination allows observation by every major PTA, providing like PSR J1645´0317, even during commissioning, SKA1-
a wealth of data for precision timing projects. Low (Braun et al. 2019) can serve as a critical tool for testing
these effects, ultimately enabling more accurate pulsar timing
5. Conclusions and deepening our understanding of interstellar propagation.
Acknowledgements. SCS acknowledges the support of a University of Galway,
This study presents long-term timing results of seven ecliptic College of Science and Engineering fellowship in supporting this work. OAJ ac-
pulsars observed with I-LOFAR over a two-to-three year period, knowledges the support of Breakthrough Listen which is managed by the Break-
focusing on the effects of dispersion due to the IISM and the through Prize Foundation. JPWV acknowledges support from the National Sci-
solar wind. A significant outcome of this research are the pre- ence Foundation (NSF) AccelNet program, award No. 2114721. The authors
cise measurements of DM using pulsar timing. The DM time se- thank Letizia Vincetti for her valuable comments on the paper. The Rosse Obser-
vatory is operated by Trinity College Dublin. I-LOFAR infrastructure has ben-
ries of four pulsars exhibit clear variations correlated with their efited from funding from Science Foundation Ireland, a predecessor of Taighde
ecliptic latitudes, highlighting the impact of the SW. We observe Éireann — Research Ireland.
that a spherically symmetric electron density model of the SW

Table 3: Spin and orbital parameters of the seven pulsars in our dataset.

Pulsar RAJ DECJ F0 F1 F2 DM Pb P9b A1

´15 ´2 ´26 ´3
(J2000) (hh:mm:ss) (dd:mm:ss) (Hz) (10 s ) (10 s ) (pc cm´3 ) (days) (10´11 ) (lt-s)
J0034´0534 00:34:21.84 ´05:34:36.95 532.7134 ´1.3484 – 13.765 1.5893 – 1.4378
J0826+2637 08:26:51.16 +26:37:11.37 1.8844 ´5.8973 ´7.2884 19.504 – – –
J1022+1001 10:22:57.98 +10:01:52.92 60.7794 ´0.1608 – 10.253 7.8051 7.1432 16.7648
J1136+1551 11:36:03.15 +15:51:18.26 0.8418 ´2.6224 ´10.472 4.848 – – –
J1300+1240 13:00:03.12 +12:40:54.41 160.8097 ´2.9448 – 10.152 25.3168 – 8.98ˆ10´6
J1645´0317 16:45:01.93 ´03:18:09.26 2.5793 738.34 – 35.741 – – –
J2145´0750 21:45:50.46 ´07:50:18.53 62.2959 ´0.1184 – 9.003 6.8389 4.2981 10.1647

Notes. Pulsar names are in J2000 format. RAJ and DECJ are Right Ascension and Declination. F0 is the spin frequency; F1 and F2 are its first
and second derivatives. DM is the dispersion measure. Pb and P9b are the orbital period and its derivative, and A1 is the projected semi-major axis.

Article number, page 10 of 11

 LOFAR: Long-term Timing Results of Ecliptic Pulsars Observed with I-LOFAR

Data Availability van Haarlem, M. P., et al., Gunst, A. W., Heald, G. 2013, A&A, 556, A2
van Straten, W. & Bailes, M. 2011, PASA, 28, 1
All the data presented are publicly available on Zenodo van Straten, W., Demorest, P., & Oslowski, S. 2012, Astronomical Research and
(Susarla 2025). This repository includes down-sampled archives, Technology, 9, 237
timing solutions, ToAs, templates and DM time series. Addi- Verbiest, J. P. W., et al., Coles, W. A., Hobbs, G. B. 2009, MNRAS, 400, 951
tional data can be provided upon request. Verbiest, J. P. W., et al., Hobbs, G., van Haasteren, R. 2016, MNRAS, 458, 1267
Verbiest, J. P. W. & Shaifullah, G. M. 2018, Classical and Quantum Gravity, 35,
133001
References Verbiest, J. P. W., et al., Porayko, N. K., Chen, S. 2024, Results in Physics, 61,
107719
Agazie, G., et al., Anumarlapudi, A., Archibald, A. M. 2024, ApJ, 966, 105 Wolszczan, A. & Frail, D. 1992, Nature, 355, 145
Agazie, G., et al., Archibald, A. M., Arzoumanian, Z. 2023, ApJ, 951, L8 Xu, H., et al., Guo, Y., Jiang, J. 2023, Research in Astronomy and Astrophysics,
Antoniadis, J., et al., Arumugam, S., Babak, S. 2023a, Astronomy &; Astro- 23, 075024
physics, 678, A50 Xu, X., et al., Zhi, Q. J., Qiao, G. J. 2021, ApJ, 917, 108
Antoniadis, J., et al., Bak Nielsen, A.-S., Bassa, C. G. 2023b, Astronomy &;
Astrophysics, 678, A48
Backer, D. C., et al., Heiles, C., Davis, M. M. 1982, Nature, 300, 615 1
Bailes, M., et al., Lorimer, D. R., Johnston, S. 1994, ApJ, 425, L41 Physics, School of Natural Sciences & Center for Astronomy, Col-
Bartel, N. 1981, A&A, 97, 384 lege of Science and Engineering, University of Galway, University
Braun, R., et al., Bourke, T., Keane, E. 2019, arXiv e-prints, arXiv:1912.12699 Road, Galway, H91TK33, Ireland.
2
Bray, J., et al., Roberts, P., Reynolds, J. 2015, Astroparticle Physics, 65, 22 School of Physics, Trinity College Dublin, College Green, Dublin 2,
Contopoulos, I., Kazanas, D., & Fendt, C. 1999, ApJ, 511, 351 D02 PN40, Ireland
Cordes, J. M. 2013, ApJ, 775, 47 3
Radio Astronomy Laboratory, University of California, Berkeley,
Cordes, J. M., Shannon, R. M., & Stinebring, D. R. 2016, The Astrophysical
CA, USA
Journal, 817, 16 4
Craft, H. D., Lovelace, R. V. E., & Sutton, J. M. 1968, IAU Circ., 2100, 1 ASTRON, The Netherlands Institute for Radio Astronomy, Oude
Donner, et al., Tiburzi, C., Osłowski, S. 2020, A&A, 644, A153 Hoogeveensedijk 4, 7991 PD Dwingeloo, The Netherlands
5
Donner, J. Y., et al., Tiburzi, C., Osłowski, S. 2019, A&A, 624, A22 Florida Space Institute, University of Central Florida, 12354 Re-
Edwards, R. T., Hobbs, G. B., & Manchester, R. N. 2006, MNRAS, 372, 1549 search Parkway, Partnership 1 Building, Suite 214, Orlando, 32826-
Gold, T. 1968, Nature, 218, 731 0650, FL, USA
Hamaker, J. P., Bregman, J. D., & Sault, R. J. 1996, A&AS, 117, 137 6
INAF - Osservatorio Astronomico di Cagliari, via della Scienza 5,
Hankins, T. H. & Rankin, J. M. 2010, AJ, 139, 168
09047 Selargius (CA), Italy
Hankins, T. H. & Rickett, B. J. 1975, Methods in Computational Physics, 14, 55
Hassall, T. E., et al., Weltevrede, P., Hessels, J. W. T. 2013, A&A, 552, A61
Helfand, D. J., Manchester, R. N., & Taylor, J. H. 1975, ApJ, 198, 661
Hellings, R. W. & Downs, G. S. 1983, ApJ, 265, L39
Hobbs, G. B., Edwards, R. T., & Manchester, R. N. 2006, MNRAS, 369, 655
Hotan, A. W., van Straten, W., & Manchester, R. N. 2004, PASA, 21, 302
Iraci, F., et al., Tiburzi, C., Verbiest, J. P. W. 2024, arXiv e-prints, Accepted in
A&A, arXiv:2410.22170
Jones, R. C. 1941, Journal of the Optical Society of America (1917-1983), 31,
488
Kazantsev, N. A. & Potapov, V. A. 2015, Astronomicheskij Tsirkulyar, 1628, 1
Keane, E. F. 2013, in IAU Symposium, Vol. 291, Neutron Stars and Pulsars:
Challenges and Opportunities after 80 years, ed. J. van Leeuwen, 295–300
Kriegel, M. & Berdermann, J. 2020, in 2020 European Navigation Conference
(ENC), 1–10
Lazarus, P., et al., Graikou, E., Caballero, R. N. 2016, MNRAS, 458, 868
Liu, K., et al., Lee, K. J., Kramer, M. 2012, MNRAS, 420, 361
LOFAR Collaboration. 2023, LOFAR2.0 White Paper v2023.1, accessed: 2024-
12-02
Lorimer, D. R. 2011, SIGPROC: Pulsar Signal Processing Programs, Astro-
physics Source Code Library, record ascl:1107.016
Lorimer, D. R. & Kramer, M. 2004, Handbook of Pulsar Astronomy (Cambridge
University Press)
Lyne, A., et al., Kramer, M., Stairs, I. 2010, Science, 329, 408
Lyne, A. G. & Manchester, R. N. 1988, MNRAS, 234, 477
McKenna, D. J., Keane, E. F., Gallagher, P. T., & McCauley, J. 2023, arXiv e-
prints, arXiv:2309.03228
Michel, F. C. & Li, H. 1999, Phys. Rep., 318, 227
Mitra, D. & Deshpande, A. A. 1999, A&A, 346, 906
Murphy, P. C., et al., McCauley, J., McKenna, D. J. 2021, A&A, 655, A16
Oswald, L., Karastergiou, A., & Johnston, S. 2019, MNRAS, 489, 310
Perera, B. B. P., et al., Demorest, P. B., Kerr, M. 2019, MNRAS, 490, 4666
Philippov, A. A., Spitkovsky, A., & Cerutti, B. 2015, ApJ, 801, L19
Rankin, J. M. 1983a, ApJ, 274, 333
Rankin, J. M. 1983b, ApJ, 274, 359
Rankin, J. M., Olszanski, T. E. E., & Wright, G. A. E. 2020, ApJ, 890, 151
Reardon, D. J., et al., Shannon, R. M., Hobbs, G. B. 2023, ApJ, 951, L6
Ruderman, M. A. & Sutherland, P. G. 1975, ApJ, 196, 51
Sobey, C., et al., Hessels, J. W. T., Weltevrede, P. 2015, MNRAS, 451, 2493
Stairs, I. H. 2003, Living Reviews in Relativity, 6, 5
Susarla, S. C. 2025, Irish-Low Frequency Array (IE613) observatory ecliptic
pulsar data release
Susarla, S. C., et al., Tiburzi, C., Keane, E. F. 2024, A&A, 692, A18
Taylor, J. H. 1992, Philosophical Transactions of the Royal Society of London
Series A, 341, 117
Tiburzi, C., et al., Bassa, C. G., Zucca, P. 2021, A&A, 647, A84
Tiburzi, C., et al., Shaifullah, G. M., Janssen, G. H. 2019, MNRAS, 487, 394
Timokhin, A. N. 2010, MNRAS, 408, L41

Article number, page 11 of 11