No Redetections of blc1 in 39 hr of Reobservation Campaigns of Proxima Centauri

From 2019 April 29 to May 4, the Breakthrough Listen (BL) project observed Proxima Centauri (ProxCen) with the CSIRO Parkes "Murriyang" telescope using the Ultra-Wideband Low (UWL) receiver (0.704–4.032 GHz), as part of a multi-wavelength campaign to monitor ProxCen for stellar flares (Zic et al. 2020). In parallel with the stellar flare analysis, BL analyzed the data with the narrowband technosignature search pipeline turboSETI (Enriquez et al. 2017) to place limits on the prevalence of radio technosignatures in the direction of ProxCen (Smith et al. 2021). This pipeline returned a narrowband signal-of-interest now known as "blc1," whose discovery and characterization are detailed in Sheikh et al. (2021). Based on the characteristics of the signal's drift rate, the occurrence of similar radio frequency interference (RFI) in the days prior to and following the detection, and the presence of RFI intermodulation products with identical morphologies across the band, Sheikh et al. (2021) concluded that blc1 was most likely human-generated RFI and not a technosignature. The BL project also performed a series of reobservations of ProxCen in 2020 and 2021 using the same instrument and receiver (Murriyang, UWL) to attempt to redetect blc1.

All data from the initial campaign and the reobservations are available at seti.berkeley.edu/blc1. In Table 1, we list all observations of ProxCen taken by BL with Murriyang UWL.

Table 1. All Observations of ProxCen Taken with the Parkes Telescope as Part of BL, with T_integration as the On-Source (ProxCen) Integration Time per Scan, ${\rm{\Delta }}t$ as the Time Resolution of the Observation, ${\rm{\Delta }}\nu$ as the Frequency Resolution of the Observation, and ${T}_{\mathrm{total},\mathrm{PC}}$ as the Total Amount of Time Spent on ProxCen

Context	UTC Date	${T}_{\mathrm{integration}}$	${\rm{\Delta }}t$	${\rm{\Delta }}\nu$	${T}_{\mathrm{total},\mathrm{PC}}$
	2019-04-29
	2019-04-30*
Flare campaign during which blc1 was detected	2019-05-01*	30 minutes	16.8 s	3.81 Hz	24 hr
	2019-05-02*
	2019-05-03
	2019-05-04*
	2020-11-19
2020 November re-observations of opportunity	2020-11-25	15 minutes	15 s	2 Hz	3 hr
	2020-11-30
	2021-01-04
2021 January re-observations of opportunity	2021-01-05	15 minutes	15 s	2 Hz	5 hr
	2021-01-07
	2021-01-13
	2021-04-29
	2021-04-30
2021 April two-year anniversary reobservation campaign	2021-05-01	30 minutes	15 s	2 Hz	21 hr
	2021-05-02
	2021-05-03

Note. Bandwidth for all observations was 704–4032 MHz. The bolded date indicates the appearance of blc1, while the dates with * indicate reappearances of a similar, though not provably identical, signal.

The observational choices made in the original flare campaign were designed to optimize flare detections, with a majority of time pointed toward ProxCen (30 minutes/pointing), with 5 minutes off-source pointings toward flux calibrator sources. The higher ${T}_{\mathrm{sys}}$ for calibrator pointings, and uneven split between on and off-source pointings, are non-ideal for standard RFI rejection techniques employed in single-dish SETI observations.

We thus changed the observing cadence and parameters in the 2020 November and 2021 January reobservations (NOV20 and JAN21) to account for some of these choices, including making the on-source and off-source scans the same length; using off-source positions devoid of bright discrete sources; and using off-sources nearer to the target. For NOV20 and JAN21, we used a 15 minutes ABACAD observing cadence; ProxCen (the A target) was observed for 15 minutes, then a nearby off-source star was observed for 15 minutes, then this cycle was repeated twice, with a different off-source star chosen each time.

For the 2021 April reobservation campaign (APR21), we chose to replicate the exact same observing script as the original 2019 observations, to maximize the chances of redetection with the same interference and/or astronomical environment. At the end of each day of observing, we visually inspected the 982 MHz region of the ProxCen scans; in the event of something resembling a redetection, we would switch to the more technosignature-optimized ABACAD cadence described above. This did not occur, and we used the replicated cadence for all five days of APR21.

In APR21, we enabled the UWL noise-adding radiometry (NAR) system, which periodically injects a signal from a calibrated noise source. This slightly increases the receiver temperature (∼10%), but the signal can be used to better calibrate the gain stability of the receiver over long timescales. For the narrowband search presented here, NAR calibration was not required.

As described in the table, all of the reobservations had a slightly higher frequency resolution (2 Hz) and slightly better time resolution (15 s) than the original flare campaign.

We again used the same software, turboSETI, to search for narrowband signals near 982.002 MHz. We ran turboSETI on all ProxCen scans from NOV20–APR21, searching for signals with a minimum drift of 0 Hz s⁻¹, a maximum absolute drift of 0.1 Hz s⁻¹ (blc1's maximum drift was 0.03 Hz s⁻¹), and a signal-to-noise ratio (S/N) of at least 10, commensurate with the original search.

We inspected the full-cadence dynamic spectrum plots of the 2196 hits that were returned by the turboSETI pipeline. All of these hits were RFI, and none resembled blc1 in drift morphology or variability. Only 3 of these hits were within ±1 kHz of blc1's original detection frequency; all three were triggers on broadband RFI that was present in APR21 and were clearly caused by a different phenomenon than blc1. Thus, there were no redetections of blc1 in the 39 hr of BL ProxCen reobservations.

We can calculate the minimum detectable flux densities for a narrowband radio technosignature search using Equation (4) from Gajjar et al. (2021):

$$\begin{eqnarray}&&{S}_{\min ,\mathrm{narrow}}=\displaystyle \frac{{\rm{S}}/{{\rm{N}}}_{\min }}{\beta }\displaystyle \frac{{S}_{\mathrm{sys}}}{\delta {\nu }_{t}}\sqrt{\displaystyle \frac{\delta \nu }{{n}_{p}{\tau }_{\mathrm{obs}}}}.\end{eqnarray} \tag{ 1 }$$

Our search had the following characteristics:

• 1.  
  ${\rm{S}}/{{\rm{N}}}_{\min }=10$ (detection threshold)
• 2.  
  β = 1 (dechirping efficiency with all drift rates below the one-to-one point, from Sheikh et al. 2019)
• 3.  
  ${S}_{\mathrm{sys}}=40$ Jy (system equivalent flux density for UWL)
• 4.  
  $\delta {\nu }_{t}=1\,\mathrm{Hz}$ (fiducial narrowband transmission frequency <3 Hz, from blc1's measured width)
• 5.  
  $\delta \nu =2$ Hz (frequency resolution)
• 6.  
  n_p  = 1 (number of polarizations)
• 7.  
  ${\tau }_{\mathrm{obs}}=[15,30]$ minutes (on-source observation time for [NOV20/JAN21, APR21] respectively).

This leads to a minimum detectable flux density ${S}_{\min }=32$ Jy for NOV20 and JAN21 and ${S}_{\min }=13.34$ Jy for APR21, corresponding to EIRPs ($\mathrm{EIRP}={S}_{\min }\times 4\pi {d}^{2}$, where d is 4.221 ly, the distance to ProxCen) of 6.4 GW and 2.6 GW respectively. APR21 was therefore more sensitive than the original observations due to its equivalent ${\tau }_{\mathrm{obs}}$ and better $\delta \nu$.

The distribution of observing hours was not uniform throughout the year, but instead in clumped sessions. The original observing campaign showed a signal present for at least 5 hr on a single day, with no definitive redetections on any of the other 5 days of the campaign, ruling out a continuous or daily periodicity. ⁸ APR21 rules out exactly annual or biennial periodicities for blc1. NOV20 and JAN21 make repetitions seem unlikely on the scales of weeks–months. We thus find no evidence that blc1 is a repeating signal on yearly timescales or shorter, but did not cover the entire periodicity space.

We can use the additional observations to further constrain the likelihood that the signal was interstellar, using expectations from diffractive interstellar scintillation (DISS) calculations. Narrowband radio signals of interstellar origin will vary from DISS if they are more distant than ∼100 pc. While they are not relevant to a source at ProxCen's distance, they would be significant for background sources and could cause a weak source to be raised above threshold or a strong source below (Cordes et al. 1997). DISS of an orbiting transmitter will correlate for an hour or more but less than a day and thus would be statistically independent between epochs given in Table 1. It is highly improbable for an intrinsically steady source be detected in one to six initial observations followed by 12 non-detections. We note that plasma lensing from a discrete cloud could produce highly episodic signal modulations, which makes DISS a highly improbable effect for these data, but not an impossible one.

Breakthrough Listen is managed by the Breakthrough Initiatives, sponsored by the Breakthrough Prize Foundation. Shane Smith and Steve Croft were supported by the National Science Foundation under the Berkeley SETI Research Center REU Site grant No. 1950897.