On the Deepest Search for Galactic Center Pulsars and an Examination of an Intriguing Millisecond Pulsar Candidate

We report results of one of the most sensitive pulsar surveys to date targeting the innermost region of the Galactic Center (GC) using the Robert C. Byrd Green Bank Telescope (GBT) at X-band (8–12GHz) using data from the Breakthrough Listen initiative. In total, we collected 9.5 hr of data covering the wider $\sim 8’$ diameter of the GC bulge, and 11 hr on the inner $1.4’$ region between 2021 May and 2023 December. We conducted a comprehensive Fourier-domain periodicity search targeting both canonical pulsars (CPs) and millisecond pulsars (MSPs), using constant and linearly changing acceleration searches to improve sensitivity to compact binaries. Assuming weak scattering, our searches reached luminosity limits of $L_{\rm min} \approx 0.14~{\rm mJykpc^{2}}$ for CPs and $L_{\rm min} \approx 0.26{\rm mJykpc^{2}}$ for MSPs – sensitive enough to detect the most luminous pulsars expected in the GC. Among 5,282 signal candidates, we identify an interesting 8.19 ms MSP candidate (DM of 2775 pc cm$^{-3}$), persistent in time and frequency across a 1-hr scan at a flux density of $S_{\rm min} \approx 0.007{\rm mJy}$. We introduce a novel randomization test for evaluating candidate significance against noise fluctuations, including signal persistence via Kolmogorov-Smirnov tests and flux-vs-DM behavior. We are unable to make a definitive claim about the candidate due to a mixed degree of confidence from these tests and, more broadly, its non-detection in subsequent observations. This deepens the ongoing missing pulsar problem in the GC, reinforcing the idea that strong scattering and/or extreme orbital dynamics may obscure pulsar signals in this region.

💡 Research Summary

This paper presents one of the most sensitive radio pulsar surveys ever conducted toward the Galactic Center (GC), using the Green Bank Telescope (GBT) at X‑band (8–12 GHz) as part of the Breakthrough Listen (BL) initiative. The authors observed two spatial scales: a wide‑area mapping of the ∼8′‑diameter GC bulge with 37 pointings (each observed three times for 5 min, totaling 9.5 hr) and a deep integration on the central pointing (A00) covering the innermost 1.4′ (≈3.3 pc) with five 1‑hr scans in Epoch 1 and three 2‑hr scans in Epoch 2, for a total of 11 hr. The observations span 2021 May to 2023 December, with the second epoch benefitting from an upgraded receiver that extended the usable band to 12.38 GHz.

Data were recorded as baseband voltages, then converted into both medium‑resolution (10 240 channels, 349.5 µs sampling) and high‑resolution (up to 53 248 channels, 43.7 µs sampling) filterbank products. After standard RFI excision (∼1.5 % for short scans, <0.1 % for long scans), the effective bandwidths were 3.69 GHz (short) and up to 4.88 GHz (long). Using PRESTO, the authors generated 500 dedispersed time series per scan covering DM = 0–5000 pc cm⁻³ with a DM step of 10 pc cm⁻³, employing 256 sub‑bands for short scans and 4096 sub‑bands for long scans to keep intra‑channel dispersive smearing below ∼20 µs. Fourier transforms were performed, red‑noise subtraction applied, and both constant‑acceleration and linear‑acceleration searches were executed to retain sensitivity to compact binaries.

The survey reached luminosity limits of Lₘᵢₙ≈0.14 mJy kpc² for canonical pulsars (CPs) and Lₘᵢₙ≈0.26 mJy kpc² for millisecond pulsars (MSPs), sufficient to detect the most luminous GC pulsars predicted by population models. In total, 5 282 candidates were identified. The most intriguing is an 8.19 ms periodic signal with a dispersion measure (DM) of 2775 pc cm⁻³, persisting throughout a 1‑hour scan at an estimated flux density of Sₘᵢₙ≈0.007 mJy. The authors term this candidate “BLPSR”.

To assess the reality of BLPSR, the paper introduces a novel randomization test. The original time–frequency data are shuffled many times to create synthetic noise realizations; each synthetic dataset is processed through the same pipeline, producing a distribution of signal‑to‑noise ratios, persistence statistics (via Kolmogorov–Smirnov tests), and flux‑versus‑DM behavior. BLPSR shows higher persistence (p≈0.02) and a non‑trivial flux‑DM correlation compared with the randomized ensemble, yet the overall statistical significance does not exceed the conventional 3σ threshold. Moreover, follow‑up observations—both at later epochs and at slightly different frequencies—failed to recover the signal.

The authors discuss several plausible explanations for the non‑detection: (1) extreme scattering in the inner parsec could intermittently broaden pulses beyond detectability; (2) the source might be in a very tight binary (orbital period ≲ days) where rapid acceleration changes smear the signal in a standard acceleration search; (3) intrinsic variability or scintillation could render the source detectable only during a narrow time window; (4) the candidate could be a statistical fluctuation masquerading as a pulsar. If genuine, the high DM implies an electron column density far exceeding predictions from the NE2001 or YMW16 Galactic electron density models, suggesting a denser ionized medium near Sgr A*. The 8 ms period would place the object firmly in the MSP regime, offering a potential probe of strong‑gravity effects if it resides in a close orbit around the supermassive black hole.

Comparing with previous GC pulsar searches at lower frequencies (4–8 GHz), this work improves sensitivity by a factor of 2–3, primarily due to the higher observing frequency (reducing ν⁻⁴ scattering), larger bandwidth, and longer integration times. Nevertheless, the lack of confirmed detections reinforces the long‑standing “missing pulsar” problem: despite theoretical expectations of 10²–10⁵ neutron stars within the central few parsecs, radio surveys have found only the magnetar SGR J1745‑2900 and a handful of distant pulsars. The results suggest that even at X‑band, scattering and/or complex orbital dynamics remain dominant obstacles.

In conclusion, the paper demonstrates that deep, high‑frequency surveys are essential for probing the GC pulsar population and introduces a rigorous statistical framework for candidate validation. The non‑confirmation of BLPSR highlights the challenges inherent in such searches and underscores the need for next‑generation facilities (e.g., SKA‑Mid, ngVLA) with greater sensitivity, broader frequency coverage, and advanced acceleration‑search algorithms. Successful detection of a pulsar in a tight orbit around Sgr A* would enable unprecedented tests of General Relativity, measurements of the black‑hole spin and quadrupole moment, and refined models of the ionized interstellar medium in the Galactic nucleus.