PSR B1706-44: The second pulsar detected by H.E.S.S. in the 10-100 GeV energy range

September 2019

Previous | Index | Next


Over fifty years ago in 1967, a new type of signal was observed for the first time by radio astronomers [1]. The signal was periodic, remarkably regular, with a peak every 1.33 seconds, and came from the same position in the sky. This was the first observation of a pulsar (CP 1919+21). Other similar signals were found soon after. Today, we know more than 2600 pulsars in radio, with a period ranging from 1.4&nsbp;millisecond to 23.5 seconds for the slowest one.

Pulsars have also been detected in other domains of the electromagnetic spectrum, mainly in gamma rays. Before 2008, there were only 7 known gamma-ray emitting pulsars. The launch of the Fermi satellite with the LAT (Large Area Telescope) onboard revolutionised the field with the discovery of more than 240 individuals in a decade, a number that keeps increasing every year.

As it was quickly understood, pulsars are one of the final stages of stellar evolution. After the death of a star (when no nuclear fusion reaction takes place anymore), the inner layers of the star collapse into a compact object. There may be different scenarios which depend mainly on the star mass: if it weights less than ∼8 times the mass of the Sun it becomes a white dwarf, the very massive stars of more than ∼30 solar masses collapse into black holes, and stars with an intermediate mass of approximately 8 to 30 solar masses evolve to neutron stars. These objects have an incredibly high density of about a billion tons per cubic centimetre on average. Due to the degeneracy pressure of fermions, matter at such densities becomes mostly made of neutrons, hence the name of neutron stars.

Pulsars are very fast spinning neutron stars with a huge magnetic field, whose N-S axis is tilted with respect to the rotation axis. The combination of these features is at the origin of the pulsed emission. A popular analogy to understand pulsars is that of the lighthouse. Due to their magnetic field, their emission is highly anisotropic (i.e. focused in a beam), and because of the star's rotation and the magnetic axis tilt, the beam moves like that of a lighthouse. The peak of the pulsed emission corresponds to the moment when the pulsar beam crosses the Earth.

The lighthouse beam analogy is, however, a simplified image and it is necessary to resort to more complex effects to account for the high-energy pulsations. According to the models, the emission region could be located inside or outside the "light cylinder": limit beyond which solid co-rotation with the pulsar is impossible as it would imply speeds greater than the speed of light. Inside the light cylinder a charged and conducting magnetosphere is thought to rotate with the pulsar. Outside the light cylinder particles escape from the magnetosphere and feed a relativistic wind with a striped shape (the "striped wind").

The radio emission is very often considered to be located near the magnetic poles. On the opposite, the gamma-ray emission cannot take place too close to the neutron star because the very strong magnetic field would interact with the gamma ray, annihilate it and create an electron/positron pair. The gamma-ray emission region could be located in the outer regions of the magnetosphere or in the striped wind.

PSR B1706-44

The pulsar B1706-44 is a 102 ms, middle-aged pulsar about 17000 years old, very bright in gamma rays (the third in the Fermi-LAT catalog of pulsars, about three times less luminous than Vela which is the brightest gamma-ray pulsar). Given its large distance of 2.3 kpc (8 times more distant than Vela), its brightness is remarkable.

It was first detected in gamma rays with the COS-B satellite in 1981 [2] and classified as an unidentified source before being recognised as a 102 ms pulsar a decade later with the Parkes radio telescope [3]. Since then, it has been detected by EGRET (Energetic Gamma Ray Experiment Telescope) [4], Chandra (in X-ray) [5], XMM-Newton (X-ray) [6], AGILE (in gamma rays) [7] and Fermi-LAT (in gamma rays) [8].

Observations and detection of PSR B1706-44 pulsations with CT5

The 28 m equivalent diameter telescope (CT5), added in 2012 to the core of the H.E.S.S. array of four 12 m diameter imaging atmospheric Cherenkov telescopes (CT1-4) has allowed to lower the detection energy threshold in monoscopic mode, bridging the gap to satellite-based gamma-ray instruments. This was first shown with the detection of the Vela pulsar in the 10 GeV to 100 GeV range [9], [10].

The results presented here use 28 hours of good quality data taken with CT5 in monoscopic mode during two campaigns in 2013 and 2015. The monoscopic analysis pipeline used for this detection is the same as the one originally developed and validated on Vela pulsar data with CT5 [10].

The H.E.S.S. light curve and that obtained with the Fermi-LAT above 15 GeV are shown in Figure 1. We define ON- and OFF-pulse intervals based on the Fermi-LAT phasogram (ON : 0.25-0.55; OFF : 0.6-0.2) to search of the signal in the CT5 data. An excess number of 7171 ± 1515 events is found with a significance level of 4.7σ (Li&Ma test [15]).

Fig. 1: PSR B1706-44 phasogram with the Fermi-LAT data > 15 GeV and with the 28 hours of HESS Mono data. The ON-pulse zone is grey, the OFF-pulse zone is hatched, and the white zone is a transition neither ON nor OFF.

A phase-resolved spectral analysis was performed using the same ON- and OFF-phase intervals as above. The fit of the data with a power law yields an index of 3.8 ± 0.4 (stat) for H.E.S.S. and 3.9 ± 0.1 for Fermi-LAT. The flux normalisation at a reference energy of 20 GeV is (4.3 ± 0.9 stat) × 10−8 and (4.4 ± 0.3 stat) × 10−8 TeV−1cm−2s−1 for H.E.S.S. and the LAT, respectively.

It was shown in [10] that the detection threshold of CT5 is close to 10 GeV (see Figure 2). Here, again, the CT5 data fit results are in full agreement with the ones obtained from the LAT above 10 GeV.

Fig. 2: PSR B1706-44 phase-resolved spectrum using Fermi-LAT data (> 100 MeV : grey, > 10 GeV : blue) and HESS-II-CT5 data (green 1σ error box). The index and the flux are compatible.

The fit of a parabola model was not attempted due to the rather low signal-to-noise ratio. The highest energy bin in the CT5 data lies in the range [54-225] GeV and displays an excess of 2782 events at a significance level of 2.5σ. Due to the large bias in energy reconstruction (see [10] and Figure 2), the average energy in this bin differs from a simple weighted mean taking into account the spectral index. An evaluation using a simulated spectrum with parameters matching those of the Fermi-LAT power law above 10 GeV predicts that 60% of events with <E>=62.7 GeV and with a dispersion of 37 GeV lie in that bin. The confidence box of the H.E.S.S. SED, shown in Figure 2 is hence limited to 62.7 GeV.


With the Fermi-LAT catalog of pulsars, it seemed like all pulsars presented a (sub-) exponential cut-off around in the GeV range. This apparently universal behaviour was questioned with the observations of the Crab pulsations by ground-based gamma-ray instruments above 100 GeV [11] and in 2015 up to ∼1 TeV [12]. In contrast to the expectation from a (sub-) exponential cut-off, the two pulses of the Crab, P1 and P2, can be fitted by a power law from GeV to TeV with no sign of a curvature.

While a majority of models expected the gamma-ray pulsations to be due to curvature radiation in a magnetospheric outer gap, it seems almost impossible to fit the Crab observations at very high energy with this type of models.

In contrast to the Crab, measurements of the Vela pulsar spectrum using H.E.S.S. and Fermi-LAT data in the 10 GeV to 100 GeV range [10], gave each, independently, a > 3σ indication in favour of curvature in the tens of GeV range. More recently, we announced the detection of hard pulsations from Vela in the 3 to beyond 7 TeV energy range [13], implying that there are two distinct spectral components in the GeV and the TeV domains.

As discussed above, the low signal-to-noise ratio on PSR B1706-44 has not allowed us to determine whether the pulsar was Vela-like or Crab-like in the H.E.S.S. data. The lack of statistics in the Fermi-LAT data prevents any firm conclusion on this point as well. The question is also still open regarding the pulsations detected above 20 GeV from the Geminga pulsar [14].

The difference observed between the Crab and Vela spectral behaviours points towards differences in physical conditions within which emission and acceleration processes come into play. Further progress in this field requires deeper observations and a larger population of sources.


[1] Bell, S.J., Hewish, A., et al. "Observation of a Rapidly Pulsating Radio Source", Nature 217 (1968), p. 709–713.
[2] Swanenburg, B. N. et al., "Second COS B catalog of high-energy gamma-ray sources", The Astrophysical Journal Letters 243 (1981), p. L69–L73.
[3] Johnston, S. et al., "A high-frequency survey of the southern Galactic plane for pulsars", MNRAS 255 (1992), p. 401–411.
[4] Arzoumanian, Z. & Thompson D. J., "Pulsed high-energy γ-rays from the radio pulsar PSR1706-44", Nature 359 (1992), p. 615–616.
[5] Gotthelf, E. V. et al. "Detection of Pulsed X-Ray Emission from PSR B1706- 44", The Astrophysical Journal 567 (2002), p. L125–L128.
[6] McGowan, K. E. et al. "XMM - Newton observations of PSR B1706- 44", Astrophys. J. 600 (2004), p. 343–350.
[7] Pittori, C. et al. "First AGILE catalog of high-confidence gamma-ray sources", Astronomy and Astrophysics 506 (2009), p. 1563–1574.
[8] Fermi LAT Collaboration. "Fermi Large Area Telescope Observations of Gamma-ray Pulsars PSR J1057-5226, J1709-4429, and J1952+3252", Astro- physical Journal 720 (2010), p. 26–40.
[9] Djannati-Ataï, A., Giavitto, G., Holler, M., et al. 2017, in American Institute of Physics Conference Series, Vol. 1792, 6th International Symposium on High Energy Gamma-Ray Astronomy, 040028
[10] H. E. S. S. Collaboration, Abdalla, H., Aharonian, F., et al. 2018, A&A, 620, A66
[11] The VERITAS Collaboration. "Detection of the Crab Pulsar with VERITAS above 100 GeV", Proceedings of the 32nd International Cosmic Ray Conference (2011).
[12] Ansoldi, S. et al. "Teraelectronvolt pulsed emission from the Crab pulsar detected by MAGIC", Astron. Astrophys. 585 (2016), A133.
[13] Djannati-Ataï, A, for the H. E. S. S. Collaboration, 29th Texas Symposium, Cape Town, SA, Dec 2017.
[14] López Moya M. et al. "Detection of the Geminga pulsar with MAGIC", ICRC 2019.
[15] Yu-Qian Ma & Ti-Pei Li. "Analysis methods for results in gamma-ray astronomy", The Astrophysical Journal 272 (1983), p. 317–324.